HAL Id: tel-01737416https://tel.archives-ouvertes.fr/tel-01737416

Submitted on 19 Mar 2018

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Use of lignocellulosic materials and 3D printing for theelaboration of conductive carbon strutures

Ying Shao

To cite this version:Ying Shao. Use of lignocellulosic materials and 3D printing for the elaboration of conductive carbonstrutures. Materials and structures in mechanics [physics.class-ph]. Université Grenoble Alpes, 2017.English. �NNT : 2017GREAI106�. �tel-01737416�

THÈSE

Pour obtenir le grade de

DOCTEUR DE LA COMMUNAUTE UNIVERSITE GRENOBLE ALPES

Spécialité : Matériaux, Mécanique, Génie Civil, Electrochimie

Arrêté ministériel : 25 mai 2016

Présentée par

Ying SHAO

Thèse dirigée par Davide BENEVENTI, Directeur de recherche, CNRS, et Didier CHAUSSY, Professeur, Grenoble INP codirigée par Philippe GROSSEAU, Directeur de recherche, Ecole des Mines de Saint-Etienne préparée au sein du Laboratoire Génie des Procédés Papetiers (LGP2) dans l'École Doctorale Ingénierie - Matériaux, Mécanique, Environnement, Energétique, Procédés, Production (I-MEP2)

Utilisation de matériaux lignocellulosiques et du procédé d’impression 3D pour élaborer des structures conductrices Thèse soutenue publiquement le 29 Septembre, 2017, devant le jury composé de :

Mme. Pascaline Pré Professeur, Ecole des Mines de Nantes, Présidente

M. Sylvain Salvador Professeur, Ecole des Mines d’Albi, Rapporteur

M. Salaheddine Skali-Lami Maître de conférences HDR, Université de Lorraine, Rapporteur

M. Davide Beneventi Directeur de recherche, CNRS, Directeur de thèse

M. Didier Chaussy Professeur, Grenoble INP, Co-directeur de thèse

M. Philippe Grosseau Directeur de recherche, Ecole des Mines de Saint-Etienne, Co-encadrant

2

3

Abbreviations

A

AM Active material

ASTM American standard test methods

B

BNC Bacterial nanocellulose

C

C Cellulose

CaLS Calcium lignosulfonate

CDCs Carbide-derived carbons

CNT Carbon nanotube

CP Cellulose powder

CVD Chemical vapor deposition

D

DAP Di-ammonium phosphate

DHP Diguanidine hydrogen phosphate

DMA Dynamic mechanical analyzer

DP Degree of polymerization

DTG/DTA Differential thermogravimetric analysis

DS Degree of sulfonation

E

EES Electrical energy storage

F

FTIR Fourier transform infrared spectroscopy

G

GDP Guanidine dihydrogen phosphate

H

4

H-B Herschel-Bulkley equations

HC Hemicellulose

HRTEM High resolution transmission electron microscopy

HTT Heat treatment temperature

K

KAS Kissinger-Akahira-Sunose equation

K-D Krieger-Dougherty equation

L

L Lignin

LIB Li-ion battery

LS Lignosulfonate

M

MCC Microcrystalline cellulose

MFC Microfibrillated cellulose

MS Mass spectrometry

N

NaLS Sodium lignosulfonate (completed name for LS)

NCC Nanocrystalline cellulose

P

PAN Polyacrylonitrile

PVD Physical vapor deposition

R

RT Room temperature

S

SEM Scanning electron microscope

SSL Spent sulphite liquors

T

5

TG/TGA Thermogravimetric analysis

X

XRD X-ray diffraction

6

Table of contents

Introduction .............................................................................................................................. 11

1 State of art ......................................................................................................................... 15

1.1 Conductive engineering carbon materials& composites: state of art ........................ 15

1.1.1 A general introduction to carbon materials ........................................................ 15

1.1.2 Engineering carbon materials ............................................................................. 16

1.1.3 Engineering carbons from biomass .................................................................... 17

1.1.4 Applications of engineering carbons in energy storage devices ........................ 21

1.2 MFC and LS as bio-carbon precursors: extraction methods, rheological and

mechanical properties ........................................................................................................... 24

1.2.1 MFC precursor ................................................................................................... 24

1.2.1.1 MFC hydrogel elaboration .......................................................................... 26

1.2.1.2 Rheological properties of MFC hydrogel ................................................... 28

1.2.1.3 Mechanical properties of MFC films and composites ................................ 30

1.2.2 NaLS precursor .................................................................................................. 31

1.2.2.1 Production and main properties of NaLS .................................................... 31

1.2.2.2 Rheological properties of NaLS solutions .................................................. 33

1.3 Pyrolysis of MFC/LS system and its kinetic modelling ............................................ 34

1.3.1 A short review of experimental thermoanalytical methods ............................... 34

1.3.2 Pyrolysis kinetics ................................................................................................ 35

1.3.2.1 Biomass pyrolysis kinetic modelling .......................................................... 35

1.3.2.2 Model-free isoconversional method for the estimation of kinetic parameters

37

1.3.3 Pyrolysis of three main components of biomass ................................................ 39

1.3.3.1 A review of thermal decomposition of cellulose, hemicellulose and lignin39

1.3.3.2 Impacting factors for biomass kinetics: environmental components and ash

content 43

1.3.3.3 Improvement of conductive carbon production during pyrolysis ............... 44

1.4 Microstructural and electrical conductivity evolution of biomass during pyrolysis . 46

2 Materials and methods ...................................................................................................... 50

2.1 Materials: ................................................................................................................... 50

2.2 Solution preparation .................................................................................................. 50

2.3 Rheological tests ........................................................................................................ 51

7

2.4 Forming methods of MFC/LS and MFC/LS/CP carbon precursors .......................... 51

2.4.1 Casting followed by air drying ........................................................................... 51

2.4.2 3D printing followed by air drying or freeze drying .......................................... 52

2.5 Pyrolysis and its thermal characterization ................................................................. 53

2.5.1 Pyrolysis ............................................................................................................. 53

2.5.2 Thermal characterization by TG/MS .................................................................. 53

2.5.3 Kinetic analysis .................................................................................................. 53

2.6 Characterization of carbonaceous chars .................................................................... 55

2.6.1 Microscopies ...................................................................................................... 55

2.6.2 Fourier transform infrared spectroscopy ............................................................ 55

2.6.3 X-ray tomography .............................................................................................. 55

2.6.4 Electrical conductivity measurement ................................................................. 55

2.6.5 Density measurements ........................................................................................ 55

2.6.6 Raman spectrometry ........................................................................................... 56

2.6.7 X-ray diffraction ................................................................................................. 56

2.6.8 Mechanical characterization ............................................................................... 57

3 Experimental results .......................................................................................................... 59

3.1 Use of MFC/LS blends as carbon precursors: impact of hydrogel rheology on 3D

printing .................................................................................................................................. 59

3.1.1 Introduction ........................................................................................................ 59

3.1.2 Pure LS/water solutions of different concentrations .......................................... 59

3.1.3 Pure MFC hydrogels of different concentrations ............................................... 60

3.1.4 MFC/LS slurries of different concentrations ..................................................... 62

3.1.5 Thixotropy of mixed slurries .............................................................................. 64

3.1.6 Relationship between the rheology and hydrogel printing ................................. 65

3.1.7 Characterization of printed aerogels before and after pyrolysis......................... 66

3.1.8 Conclusions ........................................................................................................ 68

3.2 Use of MFC/LS blends as carbon precursors: pyrolytic process characterization and

kinetics study ........................................................................................................................ 70

3.2.1 Introduction ........................................................................................................ 70

3.2.2 Morphological and structural characterization ................................................... 70

3.2.3 Thermal characterization .................................................................................... 72

3.2.4 Catalytic effect of sodium .................................................................................. 73

8

3.2.5 Pyrolysis gas analysis by mass spectrometry (MS) ........................................... 75

3.2.6 Kinetic analysis .................................................................................................. 76

3.2.6.1 Estimation of activation energies by KAS isoconversional method ........... 76

3.2.6.2 Kinetic triplets provided by model-fitting method ..................................... 78

3.2.6.3 Comparison between isoconversional and model-fitting methods ............. 79

3.2.6.4 Kinetic analysis of MFC/NaLS composites ................................................ 80

3.2.7 Conclusions ........................................................................................................ 81

3.3 Use of MFC/LS blends as carbon precursors: characterization of MFC/LS carbons 82

3.3.1 Introduction ........................................................................................................ 82

3.3.2 Morphology of MFC/LS sheet before and after pyrolysis ................................. 82

3.3.3 Chemical characterization by FTIR ................................................................... 83

3.3.4 Density evolution ............................................................................................... 84

3.3.5 Electrical conductivity evolution........................................................................ 85

3.3.6 Microstructural evolution ................................................................................... 86

3.3.6.1 Raman analysis ........................................................................................... 86

3.3.6.2 XRD analysis .............................................................................................. 89

3.3.7 Discussion .......................................................................................................... 91

3.3.7.1 Development of electrical conductivity during pyrolysis: impact of porosity

91

3.3.7.2 Development of electrical conductivity during pyrolysis: impact of

microstructures and HTTs ............................................................................................. 91

3.3.7.3 Comparison of biocarbons from various precursors in terms of electrical

property 93

3.3.8 Conclusions ........................................................................................................ 94

3.4 Optimization of the formulation of carbon precursors for 3D printing and carbon

quality ................................................................................................................................... 96

3.4.1 Introduction ........................................................................................................ 96

3.4.2 Rheological properties of MFC/LS/CP hydrogel ............................................... 96

3.4.3 Macro- and micro-morphology of MFC/LS/CP precursor and the resulting

carbons 97

3.4.4 Analysis of pyrolytic process of MFC/LS/CP composite .................................. 98

3.4.5 Characterization of MFC/LS/CP carbons ........................................................ 100

3.4.5.1 Density evolution ...................................................................................... 100

9

3.4.5.2 Microstructural evolution .......................................................................... 102

3.4.5.3 Electrical conductivity .............................................................................. 102

3.4.5.4 Mechanical properties ............................................................................... 103

3.4.6 Conclusions ...................................................................................................... 105

4 Conclusions and perspectives ......................................................................................... 107

5 References ....................................................................................................................... 110

6 Appendices ...................................................................................................................... 123

A: Elemental analysis data of the used NaLS ..................................................................... 123

B.Elemental analysis data of MFC ..................................................................................... 123

7 Published/submitted papers ............................................................................................ 125

8 French abstract ................................................................................................................ 127

10

Introduction

11

Introduction

The increasing concerns about the environmental issues are promoting the use of renewable

and environmentally-friendly materials in all domains. As the most abundant polymer

resources in nature, lignocellulosic materials, such as cellulose and lignin, have attracted

special attention nowadays for being widely used in medical, packaging, construction and

energy industries. In energy storage domain, carbon materials have been considered as ideal

candidates for electrode materials owing to their high specific surface area (>2000m2g

-1), high

electrical conductivity, good charge-discharge stability and good corrosion resistance1.

However, in order to achieve the advantages mentioned before, “noble” carbon precursors,

such as carbon nanotube and graphene, are used whose production is quite complex and

expensive. In such context, this work examined the use of biomass-derived engineering

carbons as active material in energy storage devices. Engineering carbons are commercial and

green carbons that are elaborated using a one-step pyrolysis from organic precursors2.

Biomass-derived carbons have been used as active carbon in a wide range of applications,

including soil improvement3, pollutant removal

4, greenhouse gas emission reduction

5, etc.,

thanks to their high specific surface area and low cost3,4

. Its use as active electrode material,

as was investigated in the current thesis, was inspired by Perepelkin6 who summarized an

electrical resistivity decrease of nearly 12 magnitudes for biomass precursor heat-treated at

350°C and 950°C, despite the fact that biomass has been considered as non-graphitizing

carbon precursor. More recently, some researchers have successfully elaborated electrodes

using such bio-carbons that lead to a good cycling performance7–9

. Nevertheless, no further

interpretation concerning the development of electrical conductivity during pyrolysis was

given for biomass-derived carbons.

Among various lignocellulosic materials, microfibrillated cellulose (MFC) and lignosulfonate

(LS) have caught special attention for the elaboration of carbon precursor composites10,11

.

MFC are released from cellulose fibers via mechanical treatments with potential chemical

treatment12

. Thanks to their microscopic dimensions, MFC form a network with strong

strength13

. LS is a water-soluble polymer composed of a large quantity of cross-linked

aromatic cycles14

. It is an industrial byproduct from the production of wood pulp using sulfite

pulping. The composites elaborated from MFC/LS slurries are self-standing with MFC

playing a role as mechanical reinforcement whereas LS ensures the carbon yield after

pyrolysis15,16

. However, in the frame of this work that is to use 3D printing as forming method,

it was revealed in a previous work15

that add of LS into MFC hydrogels lead to either a

disruption of the hydrogels’ stability after 3D printing due to lack of viscosity(with 10%-30%

LS), or a loss of shape definition because of the long restauration time(with 50% LS). Since

LS is an essential component to the composite due to its large contribution to the carbon

yield16

that is important to ensure the thermal stability and conductivity of as-elaborated bio-

carbons, one solution to improving the hydrogels’ printability while keeping using large

12

quantity of LS in the formulation is to add appropriate additives. In the aim of using bio-

sourced materials in this work, cellulose powder (CP) was chosen.

This thesis not only characterized the carbons obtained from MFC/LS/CP precursor in terms

of conductivity, microstructure, density as a function of pyrolysis temperature, but also

correlated 3D printing to the composite formation from its hydrogel form. Filament-based 3D

printing technic was used as the principal forming method for elaborating MFC/LS/CP carbon

precursor, owing to the fact that all pristine materials are water-processable and could form

appropriate hydrogel for printing. The interest of using 3D printing in this work mainly

consists in its flexibility to produce samples in various forms and dimensions in order to meet

different characterization purposes. For instance, as the simplest form, monolines were

printed only for tensile tests. More excitingly, electrodes could be directly printed using a

two-head printer with one head firstly printing a web-like structure then another filling the

web holes with other types of hydrogel (probably electrochemistry-strengthen components).

As a result, 3D printing not only amplifies the potential applications of MFC/LS/CP as a

printable hydrogel but also proposes an unexplored way to elaborate electrodes of next

generation.

The present work is divided in 4 experimental sections:

Section 1 studied the rheological properties of MFC/LS hydrogels and their relationship with

the printability and stability after 3D printing. This study aims at optimizing the formulation

of MFC/LS hydrogel in order to manufacture hydrogel samples that are morphologically

stable after 3D printing. Besides, drying method (air drying or freeze drying) was also

optimized according to the printing results of MFC/LS hydrogels of different concentrations.

Section 2 worked on the analysis of the thermal degradation of MFC/LS composites as well

as their decomposition kinetics. Since pyrolysis is the critical process that transforms carbon

precursors to real carbons, it is essential to investigate the reactions that occur in each

component during such process.

Section 3 characterized the bio-carbons of MFC/LS as a function of heat treatment

temperatures (HTT) since the latter plays a decisive role in carbons’ physicochemical

properties, including density, electrical conductivity, microstructure, etc.

Section 4 summarized the methods and conclusions of the 3 previous sections and examined

the use of MFC/LS/CP as carbon precursors. CP was included in the hydrogel formulation in

order to improve the printing results of MFC/LS that were summarized in section 1. The

thermal characterization of MFC/LS/CP composite was repeated as was described in section 2.

Moreover, biocarbons of MFC/LS/CP were characterized and compared to those of MFC/LS

(section 3).

The 4 sections correspond to 4 published (or submitted) papers that are listed in Chapter 7.

13

This PhD thesis was carried out principally in Laboratory of Pulp and Paper Science and

Graphic Arts (LGP2) in Grenoble, France. It is also in collaboration with SPIN (Center for

Chemical Engineering) in Saint-Etienne, France. The experimental work done during this

thesis is schematically illustrated in the figure below:

2014.10 2015.10 2016.10 2017.8

14

1. State of art

15

1 State of art

1.1 Conductive engineering carbon materials& composites: state of art

1.1.1 A general introduction to carbon materials

Carbon is one of the most abundant elements on earth and the utilization of carbon materials

began at the prehistoric age when charcoal was first introduced. From literature, no clear

definition of “carbon materials” is found. A generally accepted concept for “carbon materials”

is that they refer to materials composed predominantly of carbon, irrespective of their

structure. According to different criteria, carbon materials can be divided in different

categories. In terms of atomic arrangement, carbons, like diamond, graphite and fullerenes,

etc., form an ordered (poly-) crystalline structure while others are more-or-less disordered or

even amorphous, including engineering carbons and carbon films manufactured by PVD or

CVD2. In terms of chronological development of carbon-based materials, there are classic

carbons (period I), new carbons (period II) and nanocarbons (period III) (Fig 1.1.1). Since the

first report of fullerenes in 1985, carbon-based nanomaterials17

, including carbon nanotubes

(CNTs) and graphene, have been largely developed for the applications in catalyst supports,

adsorbents and energy storage compounds owing to their excellent physical and

electrochemical properties.

Fig 1.1.1 The chronology of the development of carbon materials (copyright of Michio et.al.18

)

The thermal expansity, electrical and mechanical properties as well as bulk density of

different forms of carbon (except nanocarbons) vary in a rather wide range, as listed in Fig

1.1.2. It appears that graphite and its intercalation compounds have a comparable electrical

conductivity to metals but much lighter weight, which explains their wide use as electrode

materials. Some of pyrolytic carbons, such as PAN-based and pitch-based carbon fibers,

compose of a quite porous structure, which may be a good candidate for solving the shape

16

stability and cycling performance problems for electrode host materials during

charging/discharging. Moreover, materials made of synthetic carbon fibers and their

composites have been currently employed in vehicles and bicycles for the sake of reducing

total weight while remaining adequate stiffness, which accounts for the decrease of fuel

consumption and manpower for driving them19

.

Fig 1.1.2 Properties of different forms of carbon (copyright of Michio et.al.18

)

1.1.2 Engineering carbon materials

The major topic in the present work is engineering carbons. Generally, they have more or less

disordered graphitic or even amorphous microstructures. In spite of excellent physical,

17

chemical and structural characters that (poly-) crystalline carbons possess, engineering carbon

materials are still the most commonly used in the industry and in everyday life owing to their

easier manufacturing method, lower costs of raw materials and production process.

Most of engineering carbons are derived from organic precursors by heat treatment in inert

atmosphere (called also carbonization process). The precursors for engineering carbons and

the corresponding products are listed in Fig 1.1.3.

Fig 1.1.3 A list of precursors for engineering carbon materials (copyright of Burchell et.al2)

1.1.3 Engineering carbons from biomass

Among the above precursors, biomass, mainly wood components, has caught special attention

these years for the synthesis of valuable carbon materials since it is an environmental friendly

and renewable resource and is highly available in nature. Bio-sourced carbonaceous materials

are mainly derived from three structural components of wood: cellulose, hemicellulose and

lignin. The content of each component in wood cell walls depends on the wood sources and

varies between 40%-45% for cellulose, 25%-35% for hemicellulose and 20%-30% for lignin20

.

Morphologically, wood cell wall is a complex laminate structure, which could be divided into

three different zones: the middle lamella, the primary cell wall and the secondary cell wall

(including s1, s2 and s3, as schematically illustrated in Fig 1.1.4). The secondary cell wall is a

major consideration for biosynthesis since it contains the main quantity of cellulose and

besides, lignin is nearly not detected in two other zones21

. The three different layers in the

secondary cell wall could be easily distinguished in an ultrastructural scale by the differences

in the orientation of cellulose microfibrils. Irrespective of layers in the secondary cell wall

zone, each wood component plays a crucial role to constructing the strength of cell wall:

cellulose works as the framework substance; hemicellulose is the matrix material cross-

linking cellulosic and non-cellulosic polymers; and lignin is the encrusting matter22

, as is

described in both longitudinal and transversal senses in Fig 1.1.5.

18

Fig 1.1.4 Illustration of cell wall structure of wood fibers (copyright Barnett et al.23

)

Fig 1.1.5 Schematic diagram for the ultrastructural arrangement of cellulose, hemicellulose

(polyoses) and lignin in the wood cell wall: a) transversal section; b) longitudinal section

(copyright of Higuchi et.al.22

)

Early in 1838, Anselme Payen declared the discovery of a “resistant fibrous solid” after a

series of acids and ammonia treatments of plant tissues. Shortly after that, cellulose was used

to name this plant constituent whose molecular formula was determined to be C6H10O5.24

As

is demonstrated in Fig 1.1.6, from the right to the left, a cellulosic fiber (essentially a part of

cell wall) is regarded as a bundle of microfibrillated cellulose (wrapped outside by HC and

possibly lignin), each of which is also an alongside package of several elementary fibrils (or

microfibrils)22

. Along each microfibril, there is an alternation of crystalline regions and

amorphous parts. The crystallinity of cellulose has been studied for years and is assigned to be

cellulose chains assembling where a strong and complex network of hydrogen bonds acts as

intra- and intermolecular connections25,26

. Besides, a disordered part has been proved to co-

exist and link with the ordered ones. Unlike crystalline regions, the nature and the mechanism

of disordered parts of cellulose microfibrils have not been clearly suggested yet27,28

.

Fundamentally, cellulose is a glucan polymer consisting of linear chains of 1,4-β-bonded

anhydro-D-glucose units, as it shows in Fig 1.1.6. Each unit is corkscrewed 180° with respect

to its neighbors, and the repeated segment is frequently taken to be a dimer of glucose, known

as cellobiose. Three hydroxyl groups are owned by one single glucose unite, ensuring its

ability to form strong and multi-scale hydrogen bonds in microfibril structure. Furthermore,

19

the frequent existence of –OH groups imparts cellulose with other characteristic properties

such as hydrophilicity, chirality, degradability and chemical variability. The number of

glucose units in one cellulose molecular chain (called also polymerization degree, DP) varies

depending on wood sources and the selection of sample locations. According to the

measurements conducted by Goring et. al.29

using light-scattering techniques, native cellulose

has an average DP ranging from 9000 to 15000 and a DP of 10000 means a linear chain

length of approximately 5 μm in wood. In terms of transversal dimensions, an elementary

fibril of about 5 nm could contain approximately 40 cellulose chains22

whereas the

microfibrillated cellulose has diameters ranging from 20 to 50 nm12

.

Fig 1.1.6 The hierarchy structure of cellulose(from the right to the left): from macroscopic

cell wall fibers to microscopic chemical molecular formula (copyright Lavoine et al.12

)

Hemicelluloses are mixtures of polysaccharides30

that occur in close association with

cellulose and lignin in the secondary cell wall of wood fiber. They differ from cellulose by the

composition of their sugar units, the length of chain and branching of the chain molecules21

.

The content of polysaccharides in hemicellulose differs from species to species but is

structurally divided into four general types31

: Xylans, Mannans, Xyloglucans and Mixed-

linkage β-glucans, all of which occur in many structural variations differing in side chain

types, localization and distribution of glycoside linkages in the main macromolecular chain.

Generally, HC has a lower molecular weight than cellulose as well as a lower chemical and

thermal stability, probably due to lack of crystallinity and lower polymerization degree21

. HCs

occur in nature having also 3 hydroxyl groups per chain unit, but unlike cellulose, they are

soluble in alkali and easily hydrolyzed by acids.

Lignin is a phenolic substance consisting of an irregular array of variously bonded hydroxyl-

and methoxyl-substituted phenylpropane units30

. Henriksson et al. firstly proposed a general

molecular formula of spruce lignin, presented in Fig 1.1.7. In contrast to cellulose which is a

linear polymer, lignin has a complex and three-dimensional structure and is formed by

polymerization of monolignols whose type depends on plant species. In coniferous wood,

lignin is built up almost exclusively by coniferyl alcohol (G-units) with a minor presence of

coumaryl alcohol (H-units), although the latter is a major constituent in compression wood

lignin. In hard woods, both G-units and sinapyl alcohol (S-units) are used as building blocks

whereas in monocotyledonous tissue, all these three alcohols are used as lignin precursors. Fig

20

1.1.8. presents the three major monolignols. In lignin network, they are mutually cross-linked

by different ether and carbon-carbon bonds20

.

Fig 1.1.7 A general formula of spruce lignin, proposed by Henriksson et al., shows the most

important inter-unit linkages between the phenylpropane units. (copyright Belgacem and

Gandini32

)

Fig 1.1.8 Schematic illustration of the three main precursors of lignin (copyright : thesis of Ida

Norberg20

)

Lignin has been seen as a promising carbon fiber precursor recently owing to its high carbon

content (over 60%, with respect to only 44% (theoretically) for cellulose precursor), aromatic

macromolecule morphology in addition to its comparably high abundance in nature (second to

cellulose). The production of carbon fibers from lignin has been reported in several works33,34

including a complete process from raw materials’ treatment to the final carbonization.

Several synthesis methods, such as pyrolytic carbonization, hydrothermal carbonization, high-

voltage arc electricity, laser ablation, have been reported for preparing amorphous

carbonaceous materials from the above biomass components with different shapes (single

fiber or fibrous web), dimensions and applications35

. Herein, pyrolysis (pyrolytic

carbonization) is chosen as the main method for synthesizing bio-sourced carbon materials

due to its relatively simple facilities and low-costs. At early stage of pyrolysis (200°C

<T<600°C), cyclization and aromatization proceed in the organic precursor with the release of

21

various organic compounds like hydrocarbons, and inorganic matters such as CO, CO2, H2O,

mainly because some of the C-C bonds are weaker than C-H bonds. Over 600°C, out-gassing

is typically H2 due to the polycondensation of aromatics. Up to 1500°C, the residues which

have “suffered” from carbonization may be called carbonaceous solids though they might still

contain hydrogen. Above 1500, graphitization begins so the residues contain more than 99%

of C which are thus called carbon materials6. The occurrence of reactions, including

cyclization, aromatization, polycondensation and carbonization, depends strongly on the

organic precursors as well as heating conditions. Sometimes these processes overlap with

each other throughout pyrolysis and therefore, the whole process from precursor to the final

carbon residues is often simply called “the carbonization”. A general schema of carbonization

process is provided by Michio et al. (Fig 1.1.9).

Fig 1.1.9 Schema of carbonization process (copyright Michio et al.18

)

1.1.4 Applications of engineering carbons in energy storage devices

With the increasing demand for portable electronic devices and electrical vehicles,

rechargeable Lithium –ion batteries (LIBs) have received special attention nowadays among

the electrical energy storage devices due to their relatively high capacity, fast charge-

discharge rate and light weight9,36

compared to conventional Ni-Cd batteries and Ni metal

hybrid cells. A comparison of currently used electrical energy storage (EES) technologies is

presented in terms of volumetric and gravimetric energy density37

(Fig 1.1.10a)) as well as

discharge time and power rating38

(Fig 1.1.10b)).

22

Fig 1.1.10 Comparison of EES technologies in terms of: a) volumetric and specific energy

density; b) discharge time and power rating

Among the EES devices, batteries, especially secondary LIBs, are found to occupy the most

application market and attract a large number of investments for fundamental and applied

research. A battery consists of numerous cells which are connected with each other in series

or in parallel in order to provide required capacity and voltage. Each cell is composed of one

anode and one cathode usually immerged into a liquid electrolyte. Li-ion battery cells employ

lithium intercalaction compounds as positive and negative electrodes materials in which the

electrical energy conversion between charge and discharge is realized by means of

electrochemical oxidation-reduction (redox) reaction. Commonly used materials for both

electrodes and the electrolyte are concluded in Fig 1.1.1139

. Carbon materials, whether

graphitic or not, are currently used as anode active material owing to a better cycling

performance and structural stability compared to Li metals or alloys. The materials for the

positive electrode are typically metal oxides of Lithium with either a layered structure, like

LiCoO2 or a tunneled structure, such as LiMn2O440,41

. The active materials are coated onto a

current collector, normally a metal foil, with a fluorinated polymer binder and an electronic

conductivity enhancer, typically a high surface area carbon black, to form a “sandwich”

structure. The electrolyte, which is liquid and made of Li salt dissolved into an organic

solvent, ensures the transfer of Li ions between the anode and the cathode in the cell. The

charge-discharge process during which the electrochemical reactions take place in both anode

and cathode is well demonstrated in Fig 1.1.12. When on charge, an external electrochemical

force is applied, which promote the decomposition (oxidation) of positive electrode material

into Li ions and compensation electrons. The Li ions then de-intercalate from cathode and

“travel” through Li salt electrolyte by ion exchange before intercalating into anode material.

Meanwhile, the compensation electrons travel in the external circuit and are received by the

anode to finish the reduction reaction. Fig 1.1.13 formularizes the reaction in each electrode

as well as the overall process. The discharge reverses the reactions and the electrons pass

through external circuit from anode to cathode to provide energy power for electronic devices.

23

Fig 1.1.11 Choice of materials for anode, cathode and electrolyte of LIBs

Fig 1.1.12 Charge-discharge process in a LIB cell

Fig 1.1.13 Reactions in both electrodes and the overall cell during charge-discharge

Based on the electrochemical reactions in a LIB cell, a promising electrode material should

conform to several requirements: i) the ability to receive and de-intercalate a large amount of

Li ions in a short time to ensure the capacity and energy density of the battery cell; ii) no

obvious change of the electrode’s volume or form during the intercalation and de-intercalation

of Li ions, which promises its structural stability and safety; iii) the previous reaction of Li

ions with the active material does not prohibit it from accepting latterly arrived Li ions, in

order to ensure the cycling performance and a long service life. Furthermore, the increasing

demand for energy consumption from the contemporary society accompanied nevertheless by

limited CO2 emission requires the development of renewable electrode materials to

manufactory more efficient and more environmental-friendly LIBs. Moreover, it would be

24

better if the complicated “sandwich” structure of conventional electrodes (active material

(AM)- binder- metallic current collector) could be reduced for the purpose of smaller and

lighter LIBs but with the ensured conductivity and efficiency.

Among the negative electrode materials listed in Fig 1.1.11, carbonaceous materials are

believed to be the most appropriate choice for several reasons. First of all, lithium insertion

into carbon leads to binary phases LiCx, which are close to equilibrium whereas its

intercalation into metal oxides yields often ternary non-equilibrium state LixMOn, which will

then decompose to electrochemical inactive matters.42

Secondly, although metallic Lithium

and Li-alloys exhibit extremely high redox potential, their corrosion problems and dendrite

formation with liquid electrolyte have become a major concern for the safety and cycle life of

LIBs. However, carbonaceous materials overcome such problems since they are usually

electrolyte-inert with a redox potential close to metallic Li. In addition, carbon materials, no

matter graphitic or not, naturally have certain controlled porosity, depending on carbon

sources and carbonization processes, which offers adequate room for the insertion de-

intercalation of Li ions without huge volumetric change. Moreover, bio-sourced carbon

materials have been studied for years, which could provide a possible solution for

manufacturing renewable LIB electrodes.

The present thesis aims at adopting pyrolytic wood derivatives (i.e. derivatives of cellulose or

lignin) as electrode host materials.

1.2 MFC and LS as bio-carbon precursors: extraction methods,

rheological and mechanical properties Within the purpose of producing porous and self-standing conductive bio-carbons, two mains

constitutional materials are used in the course of the current thesis, both of which are wood

derivatives: microfibrillated cellulose (MFC) and sodium lignosulfonate (NaLS). The reasons

for them to be chosen are multiple. Firstly, as mother materials to MFC and NaLS, cellulose is

the most abundant carbon source in nature and lignin ranks only second to it. Secondly,

synthesis methods of carbon fibers from cellulose and lignin precursor are accessible in

current literature. Furthermore, composites made of MFC and LS, regardless of forming

methods, are self-standing and functional in which MFC fibers constitute a web-like

reinforcement whereas NaLS serves as surrounding matrix and contributes mostly to the

carbon yield after pyrolysis. In terms of forming process, both MFC and NaLS are water-

processable, which makes it easier to produce MFC/LS composites by simply following a

molding /shaping step (like casting, 3D printing, etc.) and the subsequent drying step (air

drying, oven drying, or freeze drying). In the following paragraphs, not only the extraction

protocols of MFC and NaLS will be introduced, but also their rheological properties in

suspension state as well as the mechanical properties of dried MFC will be discussed.

1.2.1 MFC precursor

Currently, novel isolation methods make it possible to manufacture cellulosic material with

one dimension in nanoscale, which are referred to generically as nanocellulose. Among

nanocellulose, three principal categories are divided according to different extraction

25

processes and cellulose sources: microfibrillated cellulose (MFC), nanocrystalline cellulose

(NCC) and bacterial nanocellulose (BNC). Their different morphologies are exhibited in TEM

or SEM micrographs, as seen in Fig 1.2.1. Synonyms for each are also listed in Table 1.2.1

along with their typical sources, isolation methods as well as average dimensions. These

nanocelluloses not only inherit important properties of cellulose, such as hydrophilicity, broad

chemical-modification capacity but also possess new features as nanomaterials, like very

large surface area and high inter-fibril contact.

MFC is manufactured through mechanical homogenization with or without previous chemical

or enzymatic treatment. Such treatments aim at delaminating cellulose fibers and liberating

microfibrils (see cellulose hierarchy diagram in Fig 1.1.6).The resulting MFC is composed

alternatively of crystalline regions and amorphous ones with diameters comprising between 5

and 60 nm and lengths of several micrometres. After extraction, MFC exhibits gel-like

morphology in water, called MFC hydrogel.

Distinguishable to MFC, NCC consists of rod-like cellulose crystals with widths and lengths

of 5-70 nm and 100-250 nm, respectively. No amorphous regions are present in NCC since it

is removed from cellulose source by acid hydrolysis, often followed by ultrasonic treatment.

Owing to their crystalline structure and rod-like shape, NCC crystals present interesting

optical and liquid-crystalline properties43–45

and appear as very promising coating additives,

food packaging and gas barriers. However, NCCs have limited flexibility compared to MFC

due to lack of amorphous regions.

Bacterial nanocellulose is formed by aerobic bacteria via biosynthesis from low-molecular-

weight carbon sources like D-glucose. These bacteria, such as acetic acid bacteria of

Gluconacetobacter genus, are capable of excreting exopolysaccharide at the interface to air

when they are cultivated in common aqueous nutrient media46

. The resulting BNC hydrogel is

composed of a network of nanofibers whose diameter ranges between 20 and 100nm,

enclosing up to 99% of water. In contract to MFC and NCC that are isolated from cellulosic

sources, BNC is formed directly as a nanopolymer which contains a stable network of pure

cellulose free of lignin and hemicellulose content. The special features of BNC confer it a

variety of applications in medical implants47

, textiles, cosmetics, etc..

26

Table 1.2.1 Nanocellulose family (copyright Klemm et al.48

)

Fig 1.2.1 TEM micrographs of a) MFC49

and b) NCC50

; SEM micrograph of c) BNC48

In context of current thesis, MFC is chosen to be the main constitutional material to produce

self-standing and flexible electrode. The reasons lie on its better flexibility, abundant wood

sources and practical elaboration ways. In the following sections, manufacturing method and

rheological properties of MFC hydrogel, together with mechanical properties of MFC films

and composites, will be detailed.

1.2.1.1 MFC hydrogel elaboration

Microfibrillated cellulose is currently extracted from a number of different cellulose sources.

Since wood is the most important industrial source of cellulosic fibers, it is thus the main raw

material used to manufacture MFC. Compared to wood, non-wood sources, such as sugar beet

pulp39

, wheat straw and soy hulls51,52

, are attracting increasing interest thanks to the their

abundance in nature and the lower energy consumption during extraction processes. Cellulose

microfibrils are easier to be separated from these sources since they are extracted from

primary wall, in contrast to wood sources where cellulose fibers are presents in the secondary

wall12

.

Irrespective of cellulosic sources, MFC production starts from a cellulosic pulp which is made

from wood chips or other sources via a pulping process. As a chemical treatment, different

types of pulps are made by adding different chemical products: kraft pulp is made by using a

mixture of sodium hydroxide and sodium sulphide to dissolve non-cellulose ingredients like

lignin and hemicellulose, in which almost pure cellulose fiber could be obtained; pulping with

27

salts of sulphurous acids leads to cellulose sulphite pulp which contains more by-products,

like hemicellulose, in cellulose fibers.

Fig 1.2.2 Mechanical treatments in: a) refiner53

; b) high-pressure homogenizer54

; c) high-

pressure microfluidizer12,48

Mechanical treatments are essential to produce the gel-like MFC suspension from the original

cellulose pulp. Conventional way of mechanical treatments consists of a refining process

combined with homogenization55

. The refining is accomplished in a disk refiner (Fig 1.2.2a))

where the diluted fiber suspensions is forced through a gap between the rotor and stator disks,

which have surfaces fitted with bars and grooves, against which the fibers are subjected to

repeated cyclic stresses53

. In the following homogenization, previously refined cellulose fibers

are pumped at high pressure and forced through a spring-loaded valve assembly (Fig 1.2.2b)).

Since the valve opens and closes in rapid succession, the fibers are subjected to a huge

pressure drop with high shearing and impact forces. The combination of forces promotes a

high degree of delamination of cellulose fibers and leads to a release of microfibrillated

cellulose. The refining process is carried out prior to homogenization due to the fact that

refining could produce external fibrillation of fibers by gradually peeling off the external cell

wall layers (P and S1 layers) and exposing the S2 layer, and also cause internal fibrillation

that loosens the fiber wall, preparing the pulp fibers for subsequent homogenization

treatment56

.

After the first homogenizer being applied by Turbak et al. in 1985, recent alternatives for

mechanical treatments have been proposed: microfluidizer, grinder, cryocrusher, etc. Among

them, microfluidizer has attracted increasing attention thanks to the more uniformly sized

fibers that it could produce. In the microfluidizer, the cellulose pulp passes through thin z-

shaped chambers (whose diameters ranges between 200 and 400 µm) under high pressure, i.e.,

2070 bar. The shear rate is thus quite huge (up to 107 s

-1), resulting in the formation of very

thin cellulose nanofibrils.

a)

b)

c)

28

Table 1.2.2 Comparison of energy consumption required in mechanical treatments under

different pre-treatments

Pre-treatment Pulp type Energy required (kWh/t)

None Kraft 12 000-70 00057

None Sulphite 27 00058

Enzymatic Sulphite 150058,59

TEMPO Kraft 194060

Carboxymethylation Kraft/sulphite 50058,61

Each of previously mentioned delamination technics requires huge energy consumption. For

the purpose of producing gel-like MFC with adequate dimensions, 5-10 passes through the

homogenizer are necessary. Eriksen et al.57

determined that the electricity consumed by a

homogenizer for fibrillation of kraft pulp reaches as high as 70000kWh/t. Sulfite pulps are

easier to delaminate than kraft ones due to a high hemicellulose content and/or charge density

which facilitates cellulose disintegration. Nevertheless, 27000kWh of energy is still required

for the manufactory of one ton of MFC suspension with a high hemicellulose content58

. The

development of disintegration methods that are less energy-consumption becomes an

important topic in MFC production. The combinations of some pre-treatments and the

mechanical treatment have thus been suggested. Pre-treatments of cellulose fibers, including

enzymatic hydrolysis, TEMPO-mediated oxidation and carboxymethylation, help reducing

energy consumption in the following mechanical treatments on two major mechanisms: i) by

increasing fibers swelling in water and ii) by chemical modification on fibers’ surface.

Enzymatic pre-treatment with i.e. endoglucanse improves the swelling of fibers and makes

them less stiff and cohesive, thus decreasing the energy needed for disintegration62

. TEMPO-

mediated oxidation pre-treatment selectively converts the C6 primary hydroxylate groups of

cellulose to carbonxylate groups via the C6 aldehyde groups. As a result, nanofibrils within

the fibers separate from each other due to the repulsive forces among the ionized carboxylates,

which overwhelm the hydrogen bonds holding them together63

. As another chemical pre-

treatment, carboxymethylation increases the anionic charges in the formation of carboxyl

groups on the surface of MFC. When the charge density of pulp fibers rises, charge repulsion

leads to a drastic decrease in fiber-fiber friction and therefore less susceptibility to

flocculation as well as a decrease in clogging tendency64

. Cellulose fibers are hence easier to

delaminate. The above-mentioned pre-treatment methods are compared in Table 1.2.2 in

terms of energy consumption reduction.

1.2.1.2 Rheological properties of MFC hydrogel

MFC forms gels at low concentration in water. Photographs of MFC water dispersions

produced using two different pre-treatments were provided by Lavoine et al.12

, as seen in Fig

1.2.3. A large number of rheological studies have been conducted during the last decades on

MFC suspensions that were extracted from different sources via different disintegration

methods. The studies mostly focus on the determination of the viscosities (or shear stresses)

29

as a function of shear rate in the flow mode and the viscoelastic responses (the storage G’ and

loss moduli G”) in the oscillation mode.

Fig 1.2.3 Pictures of two different MFC suspensions furnished by FCBA, France: A) 2%

(w/w) MFC from eucalyptus, enzymatically pre-treated; B) 3% (w/w) MFC TEMPO-oxidated

from dömsjo.

It was largely reported that MFC suspensions had shear-thinning and thixotropic behaviours

during flow shear measurements65–67

, as symbolically shown in Fig 1.2.4. As it was much

observed, the viscosity of a 2% MFC hydrogel decreases as the shear rate increases in low

shear rate region (10-3

~10 s-1

) as well as in high shear rate one (100~103 s

-1). These two

inclined lines are connected by a Newtonian plateau at intermediate shear rates between 10

and 100s-1

. The shear thinning behaviour in the first region is concluded by several

authors66,68

to be the breakage of the entangled 3D network of MFC (that is flocculated in

water) when shearing is applied. Then the Newtonian plateau at intermediate shear rates

seems to be confusing whereas Karppien et al.69,70

tried to explain it by studying the floc size

distribution. They suggested that in the plateau region, the floc size increases rapidly with a

broadened size distribution, causing the collision between fibers more frequent within fiber-

rich areas, which compensates the viscosity loss driven by the applied shearing. Nevertheless,

after the plateau, floc size decreases despite an increase in shear rate, leading to the second

region of shear thinning. Furthermore, still in Fig 1.2.4, a weak thixotropic tendency of MFC

could be noticed due to the presence of a hysteresis loop during back and forth of shearing at

intermediate shear rates. Such a hysteresis loop is created since the microstructure of flocs has

been destroyed by the imposed shearing and there is not enough time for it to recover when

the shear force comes back, resulting a different in viscosity at the same level of shear rate.

Usually the loop has a small area, signifying that a short time is sufficient for rebuilding the

flocculation71

. Worth to noticing that there is a strong influence of concentration as well as pH

on MFC rheology66,69

.

30

Fig 1.2.4 Schema presenting the flow properties of a bleached sulphate MFC of different

concentrations: (1%) filled triangle; (2%) open square; (3%) open cycle; (4%) filled diamond.

The rheological studies are essential for MFC to be used as bio-ink in further 3D printing with

or without other additives. Its shear-thinning behavior together with a short relaxation time

confer it a good printability through an extrusion-type printer for the purpose of

manufacturing self-standing MFC objects in three dimensions15

. Rees et al.72

piloted the 3D

printing of MFC with two selected types of MFC and successfully obtained a printed objet

with clear grid-pattern after freeze-drying (in order to preserve the exact morphology of the

printed objet). Since then, the utilization of single or mixed MFC hydrogel as bio-ink has

attracted increasing interest.

1.2.1.3 Mechanical properties of MFC films and composites

MFC hydrogel forms films after air drying or oven drying. Many methods could convert MFC

gels into films: casting, vacuum filtration, dynamic jet, etc. Since they are analogous to

papermaking process, MFC films are also called MFC nanopapers. As water is gradually

removed from the wet web, cellulose microfibrils are getting closer and closer to each other

until that hydrogen bonds are formed within inter-fibril contact areas. A multi-layer and

porous network of MFC is established.

Compared to conventional paper, high specific tensile strength and elastic modulus have been

characterized in MFC nanopaper, according to many reports73–76

. Since mechanical

characterization depends intimately on measuring conditions as well as MFC nanopapers’

quality, such as fiber orientation, fiber sources, fiber dimensions and preparing methods, the

mechanical strengths provided by different authors are quite varied. Sehaqui et al.76

compared

NFC and traditional wood fibers in terms of mechanical strength and reported a drastic

improvement in elastic modulus by 66% as well as in tensile strength by 142%.

Some studies73,75,77

focus on the influence of wood sources and delamination process on

mechanical properties of MFC nanopaper. As it was revealed, a higher degree of pre-

treatment and mechanical treatment clearly results an improvement of tensile strength. It is

31

rather reasonable since highly refined fibers could achieve much more contact among each

other for hydrogen bonding.

In addition, many researches aim at correlating film-making methods and the final nanopapers’

quality. Sehaqui et al74,76

investigated the mechanical behaviors of MFC nanopapers made by

varied forming process and a filtration/compressed drying combined method, called Rapide-

Köthen, was recommended due to the best mechanical strength that it could provide for films

made in such a way. Since the mechanical strength is quite sensible to the fiber orientation

distribution, such a constrained method inhibits in maximum the out-plan oriented fibers,

leading to a great in-plan mechanical strength. The importance of fibril orientation was also

verified in the work of Baez et al.78

. Furthermore, since drying is the most relevant process to

the development of mechanical strength of paper due to the fact that hydrogen bonds are

formed during this period as water is evaporating, better paper quality can be achieved by

choosing appropriate drying conditions. Restrained drying induces drying stress in the paper

web where shrinkage is not allowed, resulting to higher Yong’s modulus, higher tensile

strength (by up to 40%79

) and better dimensional stability of the sheet than those dried freely.

Owing to its high-strength and elastic network together with the low density, MFC have been

used as reinforced agent in some composites or as coating materials80

. Virtanen et al.81

investigated MFC/alginate composites whose mechanical properties were significantly

improved with the increasing amount of MFC in them. Applications for such composites are

multiple : bio-degradable packaging81

, medical devices82

, electrode material83,84

, etc..

1.2.2 NaLS precursor

Lignosulfonates (LS) are formed during the sulfite pulping process by the cleavages of the α-

O-4 linkages of the randomly cross-linked network of lignin (see Fig 1.1.7), and a

sulphonation of the α-and/or γ-position of the side chains of C9 units14

. The fragments are

quite water-soluble owing to the introduction of sulphonic acid groups. The resulting

lignosulfonates are wildly used as dispersing agents, binders85

and complexing agents. LS was

chosen as another important ingredient in the context of the current thesis for multi reasons.

As a derivative of wood lignin, LS contains equally high carbon content (≈60%) but excellent

water-processable capacity despite its comparatively low-cost for manufactory. Composites

made from MFC/LS slurry are self-standing after drying and could achieve considerable

conductivity after high temperature pyrolysis thanks to the high carbon content of LS. In the

following sections, the manufacturing methods, the property in water and the rheological

properties of LS will be detailed.

1.2.2.1 Production and main properties of NaLS

Production of lignosulfonates starts from the spent sulphite liquors (SSL). The latter are

produced as a waste product from the production of cellulose pulp via a sulfite process. The

dissolved solids in SSL consist largely of lignosulfonates and the remainder of sugars and

acidic degradation products of cellulose and lignin. Various method have been used to

32

separate LS from the other constituents, including precipitation86

, dialysis87

and

chromatography88

.

The resulting lignosulfonate macromolecule forms a randomly branched polyelectrolyte (Fig

1.2.5a)). Commercial lignosulfonates normally have broad molecular weight distributions

(Mw=5000-60000 g/mol, Mw/Mn=3-12), and the degree of sulfonation (DS) varies from 0.4 to

0.7 sulfonate groups per phenylpropane unit89,90

. Moreover, since sulfonic acids are strong

acid, LS are therefore negatively charged with charge density varied depending on PH value.

A fragmented molecular structure of LS is demonstrated in Fig 1.2.5b). In the literature, the

relative content of phenylpropanoid monomers, linkage types and sulfonate groups is known

whereas their relative location and the detailed branching pattern as well as the position of

charged groups are not well known89

. Dissolved in water, LS molecules coil to form a

compact and spherical structure where the hydrophilic sulfonic acid groups are positioned at

the surface of a hydrophobic hydrocarbon core89,90

.

Fig 1.2.5 a) A schematic view of NaLS macromolecule14

(cycles indicate the charge groups:

negative for sulfonic acid groups which are attached to the molecular chain; positive for

sodium ions); b) Chemical formula of a fragment of LS molecular (copyright Fredheim et

al.89

)

Lignosulfonates have been applied for multiple industrial uses since they were commercially

available in the early 1930s, thanks to their polyelectrolyte nature. The most common

application is to work as dispersing agents to deflocculate or to stabilize the colloidal

suspensions. As a polyelectrolyte, LS adsorbs at the solid–liquid interface and infer a

repulsive force, thus reducing or eliminating the adhesion between particles in close

proximity91

. Furthermore, it is used as soil conditioner92

since it is capable of adsorbing on the

surface of soil, converting the unstructured soil into a macrostructural one. The hydro-

physical properties of soil are thus regulated with an improvement of fertility regeneration and

controlled soil erosion. LS has a cross-linked macromolecular structure, as a result, it was

reported that an improvement of plastic properties of soy protein could be observed when 30%

to 40% of LS was added93

.

a) b)

33

1.2.2.2 Rheological properties of NaLS solutions

Although NaLS is frequently incorporated into polymers in a slurry state to form the

composites for multifunction, its rheological behaviours under steady shear flow have seldom

been investigated. To our best knowledge, only Vainio et al.94

reported the Newtonian plateau

of viscosity for each LS solution with concentrations ranging from 10% to 50% (Fig 1.2.6a)).

The Newtonian behaviour of LS in solutions seems to confirm its “compact spherical

morphology”, as mentioned in the previous section.

Fig 1.2.6 a) Viscosity evolution of LS solutions as a function of shear rate; b) relative

viscosity as a function of volume fraction of LS solutions

It could be seen from Fig 1.2.6a) that the viscosity magnitude increases drastically with the

increase concentration by over 104. Such a rise of viscosity is reasonable since in higher

concentrations, LS “spheres” get closer to each other and the mutual friction is much more

important, leading to a higher shear stress necessary as shear is applied. In order to better

understand the relationship between the viscosity of LS solutions and the volume occupation

by LS in water (defined as volume fraction Φ, which could be transformed from normally

used mass concentration c via Eq 1.2.1), the Krieger-Dougherty (K-D) equation was applied,

Eq 1.2.2, which is a phenomenological model for concentrated suspension95

.

Eq 1.2.1 𝜙 =1

1+𝜌𝐿𝑆/𝜌𝑒𝑎𝑢∗(1−𝑐)/𝑐

Eq 1.2.2 𝜂𝑟 = (1 − 𝜙/𝜙𝑚)−[𝜂]𝜙𝑚

Where ρLS=1.4 g/cm3 (according to Marcus

96), ηr is the relative viscosity of each suspension

that is determined from its viscosity plateau in Fig 1.2.6a). Φm and [η] are respectively the

maximum volume fraction and the intrinsic viscosity of LS, whose values need to be

determined by Curve Fitting.

Open squares in Fig 1.2.6b), that are derived directly from the rheological experimental

results, illustrate that a viscosity divergence at volume fraction around 0.4. Curve fitting

(dotted line) using K-D equation leads to a good reproduction of experimental data and

provides the value of Φm and [η] at 0.37±0.01 and 6.8±0.3, respectively. Generally, the

maximum volume fraction for hard spheres is approximately at 0.6397

, and Φm increases with

34

polydispersity98

. Thereby, Vainio et al. proposed a monodispersed non-spherical morphology

for LS particles dissolved in solution due to the rather low Φm that they obtained.

1.3 Pyrolysis of MFC/LS system and its kinetic modelling

1.3.1 A short review of experimental thermoanalytical methods

The needs to control the carbon production during thermal decomposition of MFC/LS system

as well as to achieve a stable and homogeneous conductive network require a better

understanding of pyrolysis kinetics of the included biomass components. Experimental

thermal analysis technologies provide efficient data for pyrolysis kinetic analysis and

furthermore, for the construction of kinetic models. Such data could be obtained by

monitoring a property of the sample against the evolution of temperature or time, in a specific

atmosphere. The properties which interest thermal analyzers are listed in Table 1.3.1, along

with the corresponding measuring techniques. In the early years, thermal analysis were

conducted by isothermal methods, which means properties of samples were recorded by

repeating the experiments under isothermal conditions at different temperatures. Nowadays,

isothermal methods have generally been replaced by dynamic ones because of the narrow

temperature range that they could exploit despite the needed toilsome work, as well as the

unavoidable existence of the non-isothermal stage during the initial heating ramp. The non-

isothermal analysis technologies use modern thermobalances that allow samples to undergo a

programmed continuous temperature rise, which ensures no temperature regions are omitted.

However, the convenient dynamic methods still have shortcomings, such as the disparities of

values of kinetics parameters obtained from repeated experiments under same conditions and

the increased sensibility to experimental noise. For the purpose of overcoming these defaults,

low sample weight (<30mg) and low heating rate (<10K/min) are recommended in order to

eliminate the heat and mass transfer limitations. Moreover, the accuracy of non-isothermal

methods could be improved by collecting data from experiments performed at various heating

rates, which is generally adopted by researchers in recent kinetic studies.99–101

Table 1.3.1 A list of thermoanalytical technologies (copyright White et. al.102

)

Among the thermoanalytical technologies, thermogravimetric analysis is the most commonly

used since it records the mass decrease of solid phase due to the devolatilization during the

thermal decomposition at the imposed heating rate within the certain temperature range. Its

first-order derivative curve, -dm/dt, known as differential thermogravimetry (DTG), is

somehow more interesting for being able to provide the reaction rate evolution.

35

Thermoanalytical methods provide an excellent data base for the latter kinetic modelling.

However, it cannot tell neither the exact reactions taking place during thermal decomposition,

nor the numerous volatiles that are produced throughout the pyrolysis process, and overall, the

pyrolysis mechanism. Therefore, other analytical methods (optical, X-ray, etc.) should be

employed to provide further information concerning the chemical or structural modifications

in samples. In the present work, mass spectrometry (MS) technology is coupled with TGA for

the purpose that the former could characterize the produced volatile molecules by means of

ionization followed by the detection of these ions. The information provided by MS could, to

some extent, help determine the volatile gazes which rise from the pyrolysis of the sample,

and thus is dispensable for a better understanding of the pseudo-components in the kinetic

model.

1.3.2 Pyrolysis kinetics

1.3.2.1 Biomass pyrolysis kinetic modelling

Kinetic modelling of biomass is mainly based on its DTA data. For a better understanding of

conversion process, the term “conversion degree” is established, as is defined in Eq 1.3.1. α is

a function of the initial mass (m0), the current mass (m) and the final mass (mf) of samples, all

of which could be read from TGA data. The raison for applying α for kinetic modelling,

instead of mass loss (m/m0), consists mainly in that the definition of conversion α excludes

the part of non-reactants (ash content+ the char) during pyrolysis, by a subtraction term (m0-

mf) in the denominator in Eq 1.3.1. Moreover, the T-depended evolution of α begins at 0 and

ends at 1, with the former standing for the eve of reactions and the latter representing the final

state of reactants where they accomplish the devolatilization. Thus, α signifies more

intuitively the whole conversion routine, compared to mass loss evolution. The first-order

derivative of α, dα/dt, is called the conversion rate.

Biomass undergoes a set of quite complex reactions throughout pyrolytic decomposition.

Therefore, a hypothesis arises that pyrolysis of biomass could be considered as the sum of

several independent parallel reactions. Each reaction takes place owing to the decomposition

of a specific component of the biomass sample, called pseudo-component, since its real

chemical nature usually remains unknown. Based on such a parallel schema, the conversion or

conversion rate function of biomass is the superposition of that of each pseudo-component

multiplied by its mass fraction (𝜇𝑖), as is expressed by Eq 1.3.2 and Eq 1.3.3.

Eq 1.3.1 𝛼 ≡𝑚0−𝑚

𝑚0−𝑚𝑓

Eq 1.3.2 𝛼 = ∑ 𝜇𝑖𝛼𝑖𝑁𝑖=1

Eq 1.3.3 𝑑𝛼

𝑑𝑡= ∑ 𝜇𝑖

𝑑𝛼𝑖

𝑑𝑡

𝑁𝑖=1

Under isothermal conditions or even dynamic ones but without heat or mass transfer

limitations, the conversion rate of each pseudo-component can be influenced by two terms

(Eq 1.3.4).

36

The first one (Eq 1.3.5), known as rate constant, presents an Arrhenius dependence on

temperature, in which Ai, in s-1

, is the pre-exponential factor or the frequency factor and Ei, in

KJ/mol, is the activation energy of the corresponding reaction. For homogeneous gaseous

systems, the Arrhenius expression is proved appropriate for kinetic analysis since their energy

distribution function could be represented by Maxwell-Boltzmann equation which is the

starting point of Arrhenius equation. Therefore, the physical significance of Arrhenius

parameters can be interpreted by molecular collision theory102

. The activation energy E is

regarded as an energy barrier that must be overcome so that molecules can get enough close

to react and form products. The frequency factor, A, provides a measure of the frequency of

occurrence of the reaction103

. Thus, the rate constant, k(t), being the product of A and the

exponential term including E, yields the frequency of successful collisions104

. However,

pyrolysis of solid state materials, like biomass, is considered to be a heterogeneous chemical

reaction. There has been some criticism for the use of Arrhenius equation in solids kinetics in

a physical point of view and among them, Garn105

emphasized that this equation could only

be applicable to homogeneous reactions. In spite of that, Arrhenius-like expression has been

successfully used for biomass kinetics modelling99,101,106–108

and it has allowed for

descriptions of temperature dependence of many thermally activated solid state processes

such as nucleation and nuclei growth109

or diffusion110

since for these cases, the system must

overcome a potential energy barrier and the energy distribution along the relevant coordinate

is governed by Boltzmann statics. Besides, Galwey et. al.103

confirms the capacity of

Arrhenius equation for being applied to solid state kinetics by proving that the energy

distribution functions for both electronic and phonon energy of heterogeneous solid samples

exhibit approximately the same form as that in Maxwell-Boltzmann distribution.

The second term in the left of Eq 1.3.4 is a conversion function that represents the selected

reaction model. Table 1.3.2 lists the most common reaction models from literature. Among

them, first order reaction model (Eq 1.3.6) is widely adopted for biomass kinetics modeling

owing to its adequate adaptation to dynamic thermoexperimental results, which accounts for

being applied in the present work.

Eq 1.3.4 𝑑𝛼𝑖

𝑑𝑡= 𝑘(𝑇)𝑓(𝛼)

Eq 1.3.5 𝑘(𝑇) = 𝐴𝑖exp (−𝐸𝑖

𝑅𝑇(𝑡))

Eq 1.3.6 𝑓(𝛼) = (1 − 𝛼𝑖 )𝑛𝑖

Table 1.3.2 Expressions for the most commonly used reaction mechanisms in solid kinetics

(copyright White et. al.102

)

37

In this way, a set of differential equations could be produced, with each of them takes form of

Eq 1.3.4 to represent a single conversion process from the parallel schema. In order to obtain

a numerical solution for αi from each differential equation, two requirements must be met:

firstly, the kinetic triplet, Ai, Ei and f(α), should be given as known constants or at least, their

initial values as well as ranges of values should be defined if it is programmed in an

optimization function for purpose to provide best kinetic parameters with which the

conversion model best fits the experimental one; secondly, a powerful mathematic calculation

tool, such as Matlab, is demanded in order to provide suitable solutions with sufficient

precision for thousands of data points. More concretely, the best fitting of the constructing

model to the experimental curve, according to least squares principal, could be achieved with

reasonable initial values and value ranges of kinetic triplets and by mathematical calculation

of the differential equation set. At the same time, the optimal kinetic parameters would be

given.

1.3.2.2 Model-free isoconversional method for the estimation of kinetic parameters

Although model-based kinetic method has won a special popularity for being used to evaluate

solid state kinetics, especially for those under non-isothermal thermoanalysis conditions, the

kinetic parameters, supplied by studies (conducted under similar heating conditions) of

biomass having similar components, are sometimes inconsistent. This is mainly due to an

inappropriate choice of model function f(α). Since Arrhenius parameters are highly correlated

with the reaction model, they can be forcibly adjusted to accommodate any selected one from

Table 1.3.2 and thus result a suitable fitting to the experimental curve. For example, early

explorations111–113

for describing the single-step thermal decomposition of ZnCO3 led to a

divergence that Hüttig et al. applied a power law model with n=2/3 to derive a smaller E (38.4

kcal/mol) as compared to the result found by Bruzs that contains an activation energy value at

95 kcal/mol based on a first-order reaction model. Such a “kinetic compensation effect”114

evokes significant confusions for researchers when choosing a pertinent reaction model and

because of it, the model-independent isoconversional method attracts a lot of attention since

its empirical formula was first proposed by Kujirali et al.115

.

38

The model-free approach allows predicting Arrhenius parameters without previous

assumption of reaction mechanism. It is based on isoconversional hypothesis, that is, the

reaction rate at a given extent of conversion depends only on temperature. As mentioned

before, the first empirical isoconversional equation was proposed by Kujirali et al. to evaluate

the temperature dependence of materials decomposed under isothermal conditions. Laterly, in

non-isothermal kinetics, both differential and integral isoconversional methods have been

developed since the 1960s. Constant heating rates βj are used for most of non-isothermal

experiments, as expressed in Eq 1.3.7 (in which the subscript j represents the ordinal number

of the experiment realized at the heating rate βj).

Eq 1.3.7 𝛽𝑗 =𝑑𝑇

𝑑𝑡

Friedman suggested a differential isoconversional method (Eq 1.3.8) by combining the

heating rate βj with Eq 1.3.4 in a logarithmic form, in which k(T) is replaced by Eq 1.3.5. At a

specific extent of conversion α, the plot of ln [𝛽𝑗𝑑𝛼

𝑑𝑡] versus 1/𝑇𝑗 yields a series of parallel

straight lines, corresponding to different heating rates, whose slope is E/R.

Eq 1.3.8 ln [𝛽𝑗 (𝑑𝛼

𝑑𝑇)

𝛼,𝑗] = ln[𝐴𝑓(𝛼)] − 𝐸/𝑅𝑇𝛼,𝑗

Whereas for integral isoconversional methods, as the word “integral” indicates, they make

integration of both sides of an equation that is directly transformed from Eq 1.3.4 (with k(T)

being replaced by Eq 1.3.5 and according to Eq 1.3.7, 𝑑𝑡 =𝑑𝑇

𝛽𝑗):

Eq 1.3.9 𝑔(𝛼) = ∫𝑑𝛼

𝑓(𝛼)

𝛼

0=

𝐴

𝛽𝑗∫ exp (−

𝐸

𝑅𝑇)

𝑇𝛼

0𝑑𝑇

where Tα is the temperature at conversion α. The integral part of the rightmost side in Eq 1.3.9

is called the temperature integral. It does not have an exact analytical solution in closed form

but can be approximated by using different empirical formulas, which accounts for the variety

of integral methods. In order to simplify Eq 1.3.9 for the approximation, the term x=E/RT is

defined, so that Eq 1.3.9 is transformed as follows:

Eq 1.3.10 𝑔(𝛼) =𝐴𝐸

𝑅𝛽𝑗∫

exp (−𝑥)

𝑥2

∞

𝑥𝑑𝑥 =

𝐴𝐸

𝑅𝛽𝑗𝑝(𝑥)

One commonly used integral isoconversional approach, the Ozawa-Flynn-Wall (OFW)

method116–118

, applies an approximation formula suggested by Doyle119

, seen as the Eq 1.3.11.

The final equation yields Eq 1.3.12 where g(α) is a constant at each extent of conversion. This

equation is valid for 20<x<60. According to it, at a specific conversion degree, the

dependence of ln 𝛽𝑗on 1/𝑇𝑗 for each heating rate should be linear with the slope for such a

straight line ln 𝛽𝑗vs. 1/𝑇𝑗 should be equal to -1.0516E/R.

Eq 1.3.11 𝑝(𝑥) ≅ −0.0048𝑒−1.0516𝑥, 𝑓𝑜𝑟 20 ≤ 𝑥 ≤ 60

39

Eq 1.3.12 ln 𝛽𝑗 = ln𝐴𝐸

𝑅𝑔(𝛼)− 5.33 − 1.0516𝑥 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 − 1.0516

𝐸

𝑅𝑇𝛼,𝑗

Another widely cited integral approach is called Kissinger-Akahira-Sunose (KAS) method. It

employs a generalized empirical approximation formula (simplified as shown in Eq 1.3.13)

that yields an appropriate approximation especially for smaller E/RT111,120,121

. Substitution of

Eq 1.3.13 into Eq 1.3.10 and taking ln of both sides result the final equation of KAS method,

Eq 1.3.14. Similarly, ln(𝛽𝑗

𝑇𝑗2) has a linear relationship with 1/𝑇𝑗 for the heating rate 𝛽𝑗 so that

the activation energy could be read from the slope of the line ln(𝛽𝑗

𝑇𝑗2) vs. 1/𝑇𝑗.

Eq 1.3.13 𝑝(𝑥) = 𝑥−2𝑒−𝑥 𝑓𝑜𝑟 20 ≤ 𝑥 ≤ 50

Eq 1.3.14 ln(𝛽𝑗

𝑇𝛼,𝑗2 ) = ln(

𝐴𝑅

𝐸𝑔(𝛼)) −

𝐸

𝑅𝑇𝛼,𝑗

1.3.3 Pyrolysis of three main components of biomass

As explained in chapter 1.3.2.1, kinetics of biomass could be considered as a linear

superposition of each reactant’s single reaction along with its percentage occupation. Thus,

the study of the complicated pyrolysis process of biomass could be simplified to that of each

participating constituent. In spite of the variety of biomass species, there are only three

fundamental components: cellulose, hemicellulose and lignin, whose mass fractions might be

changing within them among different biomass types. Hence, knowledge of the kinetics of

these three components is not only necessary but dispensable for constructing any biomass

system’s kinetics.

1.3.3.1 A review of thermal decomposition of cellulose, hemicellulose and lignin

Abundant resources122–125

have proved that thermal stability of cellulose (C), hemicellulose

(HC) and lignin ( L) are quite different. Under inert atmosphere and low heating rates

(<10°C/min), HC decomposes earliest within a temperature range of 250-310°C whereas its

maximum reaction rate present at around 295°C. Then cellulose decomposes between 300 and

380°C, whose DTG peak appears at 345°C. Lignin is the most thermal- stable wood

constituent that decomposes much more slowly in a rather wide temperature range and only

exhibits a slight “peak” at around 375 °C (Fig 1.3.1). Herein, it is worth noticing that the mass

loss below 200 °C for all biomass constituents is attributed to the evaporation of moisture

content in the samples. In terms of carbonaceous residue at 800°C, lignin yields nearly 44%

of its initial mass, which is almost 2 times higher than that of cellulose (18%) and of HC

(22%).

40

Fig 1.3.1 Comparison of thermal decomposition of C, HC and L under nitrogen (copyright:

Cagnon et al.122

)

The above divergences of thermal behavior between cellulose, hemicellulose and lignin are

strongly related to their different chemical constitutions and structures. With references to

molecular structure descriptions of each of the three from chapter 1.1.3 as well as their

element compositions listed below (Table 1.3.3), it is reasonable that lignin has the highest

carbon yield together with the best thermal stability since it consists of highest carbon content

in addition to a complex network of cross-linked aromatic molecules that are difficult to

decompose during heat treatment126

. Cellulose decomposes at a well-defined intermediate

temperature range, mainly due to its homogeneous unbranched crystalline structure of linked

D-glucose units. On the other hand, Xylan (a type of HC), being also a polysaccharide, is less

thermally stable, which is attributed to its amorphous structure with many branched units that

has a low activation energy123

.

Table 1.3.3 Elementary composition of cellulose, xylan (representing hemicellulose) and kraft

lignin (copyright: Stefanidis et al.)

A large amount of studies have been conducted for kinetic modelling of the three biomass

main constituents99,100,106,127,128

. However, the kinetic triplets provided by different authors for

the same material are not always consistent, depending on experimental conditions, plant

species, the choice of method (model fitting or isoconversional) and if model fitting method

was selected, which model function is used.

Anca-couce et al.107

estimated kinetic triplets of cellulose, hemicellulose and lignin by using

model fitting method throughout two wood types (Table 1.3.4). With the choice of nth order

reaction mechanism, they successfully simulated experimental behaviors of all the three wood

components during pyrolysis in inert atmosphere (Fig 1.3.2).

41

Table 1.3.4 The kinetic triplets of cellulose, hemicellulose and lignin estimated by model

fitting using nth order reaction on two wood species (copyright: Anca-Couce et al.)

Fig 1.3.2 Fitting results when use first order reaction and kinetic data for beech-A from

Pyrolysis of NaLS

The thermal comportment of lignosulfonates has seldom been reported with respect to that of

lignin, mainly due to its poor application in biofuel domain together with the complex

reactions that they undergo during pyrolysis. One of the minority, Jakab et al129,130

.

investigated various lignosulfonates (including CaLS and NaLS) with comparison to different

lignin products by coupling TG and MS techniques. Their results reveal a total different

decomposition pathway of lignosulfonates, especially for NaLS, who presents various

reaction peaks in its DTG curve, in contrast to that of lignin whose maximum reaction rate

peak is only one (as shown in Fig 1.3.1). Furthermore, NaLS was also compared to

lignosulfonate acid (LS) and ammonium lignosulfonate (NH4LS) in terms of thermal

behaviors, showing some analogies along with differences (Fig 1.3.3). Below 600°C, all three

LSs exhibit two major mass loss peaks with the distance between them varies among NaLS,

LS and NH4LS. As ionic compounds, cation type appears to have strong effect on the thermal

stability of these lignosulfonates. The mass spectra of SO2 provides an extra evidence for

proving that the first reaction peak for all three lignosulfonates is largely reasonable to be

considered as decomposition of sulfonate group owing to similar form and consistent

42

temperature range for corresponding peaks. The second peak is difficult to be identified since

it is probably a mixture of dehydration and decarburization reactions. It is worth noticing that

dehydration reaction happens throughout the whole pyrolysis of lignosulfonates and water

remains the most important volatile product since its profiles follow roughly the feature of

DTG curves. The DTG maximum of NaLS appearing at 740°C is attributed to carbonate

decomposition.

Fig 1.3.3 Comparison of thermal behavior of NaLS −▪−▪−, NH4LS ◦◦◦ and LS ───: a) DTG

curves; b) MS profiles for SO2; c) MS profiles for H2O (copyright Jakab et al.1993)

Kinetic analysis of NaLS is even more rarely conducted than its thermogravimetric studies.

Till now, only Li et al.131

have been found to report a complete kinetic research using model-

free Kissinger approach. In his work, the activation energy of each reactant is previously

determined by Arrhenius plots. The most interesting is that the asymmetry of each reaction

peak is taken into consideration and characterized by a shape index s that is directly

associated with reaction order n. Also, the frequency factor A can be calculated for n≠1

according to an equation derived from Eq 1.3.4 (see Li et al. for more details). The Arrhenius

parameters for NaLS that they provided are listed in Table 1.3.5. Indeed, their method and

results has useful reference values for further kinetic researches of NaLS whereas the NaLS

that they used for dynamic TG experiments and the later kinetic modeling is lack of

43

universality, so that their DTG curves for modeling are much simpler than those of normal

industrial ones (NaLS used in the work of Jakab et al as well as those used in present thesis).

Consequently, there is an impending need for studying NaLS kinetics.

Table 1.3.5 Kinetic parameters of NaLS (copyright Li et al 2014)

1.3.3.2 Impacting factors for biomass kinetics: environmental components and ash content

In a majority of biomass systems, there are usually mixtures of the three main constituents of

biomass, with the proportion of each probably varying between 0 to 100%. Although it is

assumed to have no interaction between every duplet among cellulose, hemicellulose and

lignin during pyrolysis of blends, whether it is always true remains controversial since largely

different conclusions have been given from literature. Luo et al.132,133

and some other

researchers125,134

believe the existence of significant interactions between C, HC and L during

pyrolysis with investigations in terms of gas and tar production, char –formation and product

composition by using TG-FTIR, TG-MS techniques together with chemical elementary

analysis. However, if only taking their thermogravimetric or DTG results into consideration,

on which the kinetic theory was basically founded, there is no surprising disparity between the

experimental curves and the calculated ones with the latter obtained under assumption of

linear superposition of the three biomass components. Indeed, influences might exist with the

addition of other wood component(s) into the original single one during pyrolysis, in a way of

promoting or inhibiting the production of some gases or chars, but if equilibrium is achieved

between promoted and inhibited reactions, the total reaction rate of the blends could remain

equal to the sum of each, same tendency as is suggested by their own results. Furthermore,

instead of concluding the existence of interactions between wood components, Yang et al.124

attribute such a slight shift of reaction rate peak towards higher temperature in C/HC/L blends

to the mass and heat transfer limitations that are caused by the addition of other wood

component(s) with different particle size(s) and chemical nature. Therefore, in a point of view

of kinetic studies, it is reasonable to suppose that no interaction between cellulose,

hemicellulose and lignin occurs throughout pyrolysis, which is consistent with mainstream

researches in the domain of biomass kinetics135–137

.

Another discussable factor that could modify kinetic triplets of biomass is the ash content.

Different extraction methods are used to produce industrial “pure” wood components and

44

supplementary acid or alkali treatments are applied to enable raw materials to meet strict

dimensional and structural requirements for scientific research purpose. Consequently, the

raw materials used for pyrolysis possess more or less inorganic mineral content, called also

ash content. As indicated in Table 1.3.3, industry-available cellulose, hemicellulose and lignin

have tiny amount of ash whereas commercial NaLS, on the other hand, contains a quite high

mineral percentage (as high as around 20%). Generally speaking, the main elemental

constituents of biomass minerals are Si, Ca, K, Na and Mg with smaller amounts of S, P, Fe,

Mn and Al, all of which occur as oxides, silicates, carbonates, sulfates, chlorides and

phosphates in ash content138

. Both Müller-Hagedorn et al.128

and Nassar139

found that alkaline

salts in biomass, whether added or innate, act to lower the apparent activation energy of

thermal reactions and promote the formation of char. Furthermore, Varhegyi et al.140

compared treated sugarcane bagasse samples with diluted inorganic salt solutions (such as

MgCl2, NaCl, FeSO4, and ZnCl2) and non-treated ones in terms of thermal comportment and

gas/char production by MS, revealing that the increasing char production in treated samples

(except for the case of MgCl2 in which no difference was identified between treated and non-

treated samples) could be attributed to an alteration of reaction pathways by theses salts. The

MS intensities of all treated samples were lower than those of untreated ones, suggesting the

presence of inorganic additives suppresses the secondary cracking of high molecular weight

primary products. In addition to catalytic effects imposed by cation contents in ash, anions

were also observed to influence the pyrolysis temperature with an impacting order:

bicarbonates<sulfates<chlorides128

.

1.3.3.3 Improvement of conductive carbon production during pyrolysis

The application of carbonaceous biomass into electro-chemical devices requires them to have

good electrical conductivity in addition to appropriate mechanical strength. In order to

guarantee both properties, the pyrolytic process that biomass undergo should be well

controlled with the purpose of promoting the dehydration reaction and char formation

whereas inhibiting carbonaceous volatiles production. Numerous solutions have been

suggested to efficiently improve the quality of carbonized biomass, which could be divided

into several aspects: i) high pyrolysis temperature; ii) controlled heating rate; iii) pre-

treatment of sample with various impregnating agents.

Perepelkin6 suggested a relationship between heat treatment temperatures (HTT) and

electrical resistivity of different biomass precursors (Fig 1.3.4). In his chart, it is clearly

illustrated that, regardless of wood precursors, an increase of HTT declines observably

sample’s resistivity, indicating a rise of electrical conductivity in the other hand. Pyrolysis up

to 750°C allow to convert all types of biomass into conducting agents, which is also in

agreement with the fact that the higher heat treatment temperature is, the purer carbon

material is obtained.

45

Fig 1.3.4 a) Significance of temperature on final electrical resistivity of carbon biomass: 1)

and 2) hemicellulose precursor; 3)charcoal; 4)lignin carbon (copyright Perepelkin 2002); b)

influence of heating rate on char yield during cellulose pyrolysis (copyright: Brunner et al.,

1980)

Brunner et al.141

investigated in detail the importance of heating rate on char yield and char

properties in cellulose pyrolysis. As it is illustrated in Fig 1.3.4b), a change of heating rate

from 70 to 0.03 °C/min results a considerable increase in char yield from 11% to 28% at the

end of pyrolysis at 900°C. This is due to a prolongation of dehydration reaction at low

temperature (<240°C), which leads also to thermally more stable char with a low oxygen

content. With examination of char properties, it is concluded that low heating rates help

likewise to yield highly porous but dense chars.

Pre-treatment of biomass with additives increase its reaction rate during pyrolysis. Hence,

additives play a dispensable role in biomass pyrolysis by means of enhancing reaction kinetics

by cracking higher molecular weight compounds into lighter hydrocarbon products142

. A great

number of additives of various types have been suggested to improve the quality of biomass

char according to different mechanisms. Flame retardants have been widely used since they

could facility and stabilize the pyrolysis process of biomass143

. Gao et al.144

investigated

several of them, including guanidine dihydrogen phosphate (GDP) and diguanidine hydrogen

phosphate (DHP), achieving a rise of 33% on carbon yield. Zeng et al.145

tried several

phosphate and ammonium salts, proving the usefulness of di-ammonium phosphate (DAP) on

improvement of carbon yield along with a high specific area. Besides, Lysenko et al146

demonstrated that water-soluble organosilicon, whether alone or mixed with other ammonium

additives, helped increasing carbon yield to an important extent and improving

simultaneously mechanical resistivity of carbon fibers. Moreover, impregnation of biomass

samples in a diluted sulfuric acid solution147

before heating treatment or conducting the

pyrolysis process of biomass in HCl atmosphere148

helps increase the carbon yield to 38%

together with a considerably restrained shrinkage observed in cellulose fibers under SEM. It is

due to the fact that strong acids work as dehydration agents which keep “absorbing” H2O

molecules throughout heating. Other examples of additives are alkaline-sodium compounds,

a) b)

46

such as NaOH, Na2CO3 and Na2SiO3, which decrease gaseous products like CH4 and CO2 and

favor H2 formation mostly149

. A lot of more additives could be found in literature and the

selection of pyrolysis catalysts should also consider the resulted impurity that they bring into

the final carbons.

1.4 Microstructural and electrical conductivity evolution of biomass

during pyrolysis Biochars which are derived from lignocellulosic materials via a pyrolysis process have been

used as environmentally-friendly carbon materials in a wide range of applications, including

soil improvement3, pollutant removal

4, greenhouse gas emission reduction

5, etc., owing to

their highly porous structure and the resulting high surface area3,4

. Despite the fact that

biomass has been considered as non-graphitizing carbon precursor, Perepelkin6 summarized

an electrical resistivity decrease of nearly 12 orders of magnitude for biomass precursor heat

treated at 350°C and 950°C, leading to a potential use of biochars as “active” electrode

materials in energy storage devices. Although outstanding cycling performances have been

achieved on such biomass-derived electrodes7–9

, little interpretation has been made for the

electrical conductivity development of biomass during pyrolysis.

During pyrolysis, carbon precursors suffer from molecular decomposition and condensation to

form carbonaceous chars with a carbon content up to ca. 80%150

(800°C). As intrinsic

property, the electrical resistivity of a carbon material, with the latter seen as a packed bed of

carbon particles, should account for both factors: the intra-particle resistance and the inter-

particle (contact) resistance1,151

. Both factors are strongly temperature-depended152,153

and

could thus be characterized by the microstructural evolution and the percolation model as a

function of heat treatment temperature (HTT).

McDonald-Wharry et al.150

summarized previous descriptions of the microstructure of non-

graphitizing carbons by establishing a model called “distorted graphene triad” (Fig 1.4.1).

According to it, non-graphitizing carbons are composed of 3 distinguish microstructures (Fig

1.4.2): i) regular graphite-like domains formed by average 3 graphene layers; ii) distorted

graphite regions (mainly due to the existence of oxygen functional groups) which forms

curved cross linking between regular graphite domains; iii) micropores with a diameter

similar to the length of regular domains. Pré et al.154,155

studied the nanostructure in activated

carbons and carbide-derived carbons (CDCs) using high resolution transmission electron

microscopy (HRTEM) which provides information about the shape, size and orientation of the

defective graphene sheets.

Rhim et al.10

studied the electrical property development throughout the pyrolysis of

microcrystalline cellulose (MCC) and suggested that the increasing HTT leads to the

continuous growth of conductive carbon clusters (regular and distorted graphite regions as

mentioned above) and upon a percolation threshold at HTT between 600 and 610°C with a

conductive phase volume fraction of 0.39, conductive phrases begin to make contact so that

47

the DC conductivity becomes detectable in MCC chars. Kercher et al.156

deduced similar

conclusions by studying the fiberboard carbonization in terms of electrical property.

Fig 1.4.1 The structure of a “distorted graphene triad” (copyright McDonald-Wharry et al.150

).

Interpretation of features: (1) persistent oxygen functional groups becoming trapped by

aromatic growth, (2) deep fjord regions within the graphene-like structure, (3) sites for

aromatic condensation (fjord closure) by elimination of H2 or two radicals, (4) formation of

conjugated cross-links that extend the π system across the structure, (5) domains of regular

graphene structure that have an approximate diameter of 1−2 nm, (6) carbonyl groups under

steric strain that could induce/fix curvature and non-planarity, and (7) bays and/or K regions,

which are more reactive, that could be favored sites for epitaxial growth to extend graphene-

like domains or form additional cross-linking.

Fig 1.4.2 Illustration for the stacking of “distorted graphene triad” (copyright McDonald-

Wharry et al.150

)

Among various lignocellulosic materials, microfibrillated cellulose (MFC) and lignosulfonate

(LS) have caught special attention for the elaboration of carbon precursor composites10,11

. The

composites elaborated from MFC/LS slurries are self-standing with MFC playing a role as

48

mechanical reinforcement whereas LS ensures the carbon yield after pyrolysis15,157

. In

addition to the thermal stability158

, MFC/LS derived carbons could acquire adequate

porosity159

and potential electrical property with regard to their comparatively low density160

,

as most of biochars do, make them a promising materials for electrodes in energy storage

devices1.

Since HTT is the predominant factor for the determination of chars’ chemical and physical

properties, the characterization of chars derived from MFC/LS precursor in terms of

morphology, electricity measurements, density and microstructure evolution needs to be

performed as a function of HTT, which, to the best of authors’ knowledge, has not been

investigated.

49

2. Materials and methods

50

2 Materials and methods

2.1 Materials: MFC, in the form of a 2% (w/w) hydrogel, was provided by FCBA (Saint Martin d’Heres,

France). It was produced from bleached hardwood (birch) kraft pulp via a mechano-enzymatic

protocol along with subsequent homogenization at high pressure.

Fig 2.1.1 Morphology of the used MFC under a) optical microscopy and b) scanning electron

microscopy

Sodium lignosulfonate (NaLS) was purchased from Carl Roth GmbH + Co. KG (France).It is

in the form of a brown powder with sodium content reaching 9%. More detailed elemental

analysis of NaLS is shown in Appendix A.

High purity cellulose powder (CP) from cotton was purchased from Sigma-Aldrich (France)

with an average particle size of 50 µm.

2.2 Solution preparation Single or mixed MFC and LS solutions of various concentrations were elaborated as follow:

Single LS solutions were prepared by simply dissolving LS of different quantities into

deionized water. The following concentrations (w/w of water) were made: 20%, 30%, 40%,

50%, 52.5% and 55%, corresponding to volume fractions ranging from 0.15 to 0.47.

MFC hydrogels (0.5%, 1%, 11.4%, w/w of water) were prepared either by dilution of the

pristine 2% with deionized water (for 0.5% and 1% ones) or by centrifugation (for 11.4%

ones).

MFC/LS mixed slurries were elaborated by adding different quantities of LS into the

corresponding MFC hydrogel. Three series of mixed slurries were prepared: i) 0.5% MFC

series which contain 0.5% of MFC and 20%-50% of LS; ii) 1% MFC series with 1% of MFC

and 20%-50% of LS; iii) 2% MFC series with 2% of MFC and 20%-50% of LS.

MFC (dry matter), LS powder and cellulose powder were 1: 49: 14 mixed for 3D printing

purpose.

a) b)

51

2.3 Rheological tests Rheological tests were conducted in two modes: simple shear mode for viscosity

measurements and thixotropic mode for thixotropic tests. Regardless of modes, all rheological

measurements were performed by using a rotational physical MCR 301 rheometer (Anton

Paar) in a plate-cone configuration. A cone with 50 mm diameter and 1° angle was used and

the gap was set to 1 mm. A transparent cover was used to prevent water evaporation during

measurements. Temperature of the plate was maintained at 23 °C.

Viscosity measurements were carried out for all suspensions by repeating several cycles with

shear rate ranging between 10-3

and 103 s

-1 with about 10 min relaxing time between each

cycle. Four measuring points were set for decay with 10 s between each measuring point.

Thixotropic measurements were carried out just for slurries with 2% MFC. The shear rate was

maintained at 1000 s-1

for 20 s before a sudden drop to 0.1s-1

, as was schematically described

in Fig 2.3.1. Samples’ viscosity and stress responses were recorded as a function of time. In

order to get rid of inertial instabilities, data were recorded every 1 s after the step-down

variation.

Fig 2.3.1 Descriptive scheme for thixotropic measurements

2.4 Forming methods of MFC/LS and MFC/LS/CP carbon precursors

2.4.1 Casting followed by air drying

It consists the simplest way to elaborate MFC/LS composites. Mixed slurries were poured in

Teflon molds and a stainless ruler was used to spread out the slurry within the mold as well as

to eliminate the extra slurry in order to smooth its surface. After drying in ambient conditions,

film-like composites were obtained (Fig 2.4.1).

52

Fig 2.4.1 An example of film-making by casting

2.4.2 3D printing followed by air drying or freeze drying

A commercial 3D printer (Model 3, Seraph Robotics) was used to test the processability of 2%

MFC suspensions with LS content ranging between 0 and 50%. Two series of square cuboids

(LxWxH: 2x2x1 cm) were printed using a 0.96 mm syringe needle and a printing speed of 35

mm s-1

(corresponding to a shear rate in the needle tip of ca. 300 s-1

). The first series was

dried at room temperature, whereas the second series, after 2h relaxation at room temperature,

was slowly frozen in a refrigerator (-12°C). Frozen samples were then freeze dried in a freeze-

dryer (Martin Christ Gefrierstrocknungsanlagen GmbH).

MFC/LS/CP mixtures were printed using a 3D printer (Leapfrog, Creatr HS model). The

triple slurry was stored in a syringe with the plunger being pushed under a steady pressure

(about 1bar) from a vacuum pump. The extruded slurry then passed through a rotating screw

before being printed via a 0.5 mm needle (Fig 2.4.2a)). The printing speed, the width of

filament as well as the layer height were set to be 550mm/min, 0.6mm and 0.33mm,

respectively, in order to obtain a good morphological definition. Samples printed in various

forms and patterns were then air-dried at ambient temperature.

Single lines of MFC/LS/CP mixture were printed using a 1mm needle (Fig 2.4.2b)) only for

bending test purpose. They were then air-dried in ambient temperature.

Fig 2.4.2 3D printer (Leapfrog) with an extruder of a) 1mm and b) 0.5mm diameter

a) b)

53

2.5 Pyrolysis and its thermal characterization

2.5.1 Pyrolysis

Pyrolysis of dried composites was conducted in a tubular oven (Carbolite®, type 3216) under

nitrogen flux (Fig 2.5.1). The heating schedule was listed in Table 2.5.1, with reference to

Kercher at al.156

.

Fig 2.5.1 The used tubular oven

Table 2.5.1 Heat treatment program

T range (°C) Heating rate (°C/min) Dwell time at final temperature

(min)

Ambient T~150 0.5 60

150~400 0.2 -

400~600 0.5 -

600~target T (600~1000) 1 10

2.5.2 Thermal characterization by TG/MS

Thermogravimetric experiments are performed in a SETARAM TG92 thermobalance under

helium atmosphere. A mass spectrometer (OmniStar, PFEIFFER) is coupled to the TG

analyzer through a transfer line heated at 180°C. Paper-formed samples are previously cut

into tiny fragments and the mass of each sample is controlled to be less than 12 mg in order to

avoid mass and heat transfer limitations during pyrolysis. TG data are recorded for all samples

in a temperature range from 150°C to 800°C at heating rate of 5°C/min. Extra runs at heating

rates of 2.5, 10 and 15°C/min were performed only on pure MFC and NaLS samples for the

isoconversional kinetic analysis. Ionized gas molecules of m/z ratios between 1 and 200 are

scanned by the MS. However, only important molecules were selected for the interpretation.

The recorded intensities are normalized to those of helium and to the initial sample mass.

Each TG-MS experiment is duplicated in order to ensure the reproducibility of the results.

2.5.3 Kinetic analysis

Model-based approach: the conversion degree of a matter is defined in Eq 2.5.1 as a

function of the initial mass (m0), the final mass (mf) and the current mass (m), derived from

the TG data of the sample. The total conversion (or conversion rate) of biomass is the sum of

54

those of its pseudo-components multiplied by their proportions (Eq 2.5.2) with N representing

the number of identified pseudo-components. As shown in Eq 2.5.3, the conversion rate for

each pseudo-component, dαi/dt, depends on: i) the kinetic constant k(T) in which Ai, Ei and R

are the pre-exponential factor, the activation energy and the gas constant (Eq 2.5.4),

respectively; ii) the reaction model f(α) for which the first order reaction model is chosen as

shown in Eq 2.5.5, with ni=1 (i=1~N). Conversion rate equations for all pseudo-components,

which are first order differential equations, are numerically solved using the ode23s Matlab

ODE solver. Arrhenius parameters, Ai and Ei, as well as the proportions of the different

pseudo-components µi, are determined by least square minimization of the objective function

defined in Eq 2.5.6, with L being the number of experimental points. The relative deviation

expressed in Eq 2.5.7 is considered to assess the quality of the fit.

Eq 2.5.1 𝛼 = (𝑚0 − 𝑚)/(𝑚0 − 𝑚𝑓)

Eq 2.5.2 a) &b) 𝛼 = ∑ 𝜇𝑖𝛼𝑖𝑁𝑖=1 , 𝑑𝛼/𝑑𝑡 = ∑ 𝜇𝑖 ∗ (𝑑𝛼𝑖/𝑑𝑡)𝑁

𝑖=1

Eq 2.5.3 𝑑𝛼𝑖/𝑑𝑡 = 𝑘(𝑇)𝑓(𝛼𝑖)

Eq 2.5.4 𝑘(𝑇) = 𝐴𝑖exp (−𝐸𝑖/𝑅𝑇(𝑡))

Eq 2.5.5 𝑓(𝛼𝑖) = (1 − 𝛼𝑖)𝑛𝑖

Eq 2.5.6 𝑂𝐹 = √∑ (𝛼𝑚𝑜𝑑𝑒𝑙 − 𝛼𝑒𝑥𝑝)2𝐿𝑙=1

Eq 2.5.7 𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 = 100|𝑚𝑒𝑥𝑝 − 𝑚𝑚𝑜𝑑𝑒𝑙|/𝑚𝑒𝑥𝑝

Isoconversional approach: constant heating rates βj are expressed in Eq 2.5.8 (in which the

subscript j represents the ordinal number of the experiment realized at heating rate βj).

Isoconversional method makes integration of both sides of an equation that is directly

transformed from Eq 2.5.3 (with k(T) being replaced by Eq 2.5.4 and according to Eq 2.5.8,

dt=dT/βj) where Tα is the temperature at the conversion level α. The integral part of the

rightmost side in Eq 2.5.9 is the so-called “temperature integral”. Kissinger-Akahira-Sunose

(KAS) method employs a generalized empirical approximation formula for the temperature

integral111,120,121

with the final equation expressed in Eq 2.5.10.

Eq 2.5.8 𝛽𝑗 = 𝑑𝑇/𝑑𝑡

Eq 2.5.9 𝑔(𝛼) = ∫ 𝑑𝛼/𝑓(𝛼)𝛼

0= 𝐴/𝛽𝑗 ∫ exp (−𝐸/𝑅𝑇)

𝑇𝛼

0𝑑𝑇

Eq 2.5.10 ln(𝛽𝑗/𝑇𝛼,𝑗2 ) = ln(𝐴𝑅/𝐸𝑔(𝛼)) − 𝐸/𝑅𝑇𝛼,𝑗

Where Tα,j represents the temperature corresponding to the degree of conversion α for a given

sample pyrolyzed at a specific heating rate ‘j’. The plot of 𝑙𝑛(𝛽𝑗/𝑇𝛼,𝑗2 ) versus 1/Tj for each

conversion level α yields a straight line with the slope and the interception with vertical axis

corresponding respectively to (-E/R) and ln(AR/Eg(α)). The kinetic parameters (activation

energy E and the pre-exponential factor A) could be deduced directly from the Arrhenius plots.

55

2.6 Characterization of carbonaceous chars

2.6.1 Microscopies

Morphological characterization was performed on samples, in both surface and sections, using

a scanning electron microscope (FEI-Quanta 2000, ESEMTM

).

2.6.2 Fourier transform infrared spectroscopy

Chemical structure characterization is conducted with a Perkin Elmer FT-IR spectrometer

(Perkin Elmer, USA). Each sample was grounded and then mixed with KBr powder (IR

spectrometer grade) with a ratio of 1:125 in order to make testing pellets. The spectra were

recorded in the wave number range of 4000 to 600 cm-1

with a resolution of 2 or 4 cm-1

and

an accumulation of 32 scans per analysis. Absorption spectra were obtained and corrected

with the environmental spectrum.

2.6.3 X-ray tomography

X-ray tomography (Nanotom 180, Phoenix X-Ray) was used to observe the internal structure

of printed composites before and after pyrolysis. Images of every transversal section along the

length of samples were recorded in order to restore the 3D structure of samples.

2.6.4 Electrical conductivity measurement

Electrical conductivity of carbonized samples was measured either by a two point ohmmeter

(Fluke) or by the four-point probes system (Standa 019759), depending on the shape of

samples.

Thin samples (thickness <1mm) were measured using a four-point probe system (Standa

019759) by following ASTM C611-98 (2005) and ASTM F84-98. The measuring current was

generated by an impendence analyzer (Jandel, model RM3). The thicknesses of a sample were

measured by a micro-comparator within the zone where the conductivity was measured in

order to calculate the average thickness. The correction factor used to compensate the finite

dimensions of each sample was from the work of Smits et al.161

. 5 samples were measured per

HTT (heat treatment temperature).

Thick objects were measured by an ohmmeter along the z axis after deposition of zinc

electrodes with a diameter of ca.7 mm on its surfaces. The electric conductivity was then

calculated with the Ohm law using the sample average thickness, the apparent contact area of

the zinc electrodes and the cross sectional area of the whole square face in order to get an

estimation of maximum and minimum conductivity values.

2.6.5 Density measurements

Skeletal density (true density) was measured using a gas displacement pycnometry system

(AccuPyc 1330, Micromeritics, USA). Helium was used as displacement gas thanks to its

extreme inertness (avoid being adsorbed) and the small molecule size (about 0.2 nm)162

, thus

it can penetrate even the tiny pores in the sample.

56

2.6.6 Raman spectrometry

Raman spectra of carbonized chars were recorded using a Renishaw in-via Raman micro-

spectrometer. A 100mW, 785nm laser was passed through a 1% transmission filter before

projecting a measurement spot on samples with a diameter of approximately 1.5 µm. A 50x

objective was used and frequencies between 800 and 2000 cm-1

were scanned in a

synchroscan mode for each spectrum. 10 spectra were collected for each char sample

representing 10 different measured positions.

The average of 10 spectra was calculated to represent the Raman pattern of one sample. The

spectrum of each sample was then smoothed and normalized to its own G-band position

(around 1600cm-1

). As presented in Fig 2.6.1, 5 Gaussian peaks (at around 1200, 1330, 1500,

1585, 1750 cm-1

, respectively) were assigned based on data from previously published

articles163–168

by considering the shape of the spectra and for the sake of highlighting the

characteristic bands.

Fig 2.6.1 Raman band assignment used in this work

2.6.7 X-ray diffraction

X-ray diffraction (XRD) was used for phase and crystal structure identification. Each char, in

the form of plate thin film, was placed on the sample holder and levelled to obtain total and

uniform X-ray exposure. The samples were analyzed using an X-ray diffractometer (X’pert

Pro MPD, PANalytical) at room temperature (RT) with a monochromatic CuKα radiation

source (λ = 0.154 nm) in the step-scan mode with a 2θ angle ranging from 5° to 70° with a

step of 0.067° and scanning time of 6.0 min.

Diffractograms of all samples were smoothed, normalized to 2θ=70° and linear-subtracted as

described in the work156

. Average crystallite size, along c- and a- axis was determined using

Scherrer equation156,169,170

Eq 2.6.1 &Eq 2.6.2:

57

Eq 2.6.1 𝐿𝑐 = 0.89𝜆/𝐵𝑐𝑐𝑜𝑠𝜃𝑐

Eq 2.6.2 𝐿𝑎 = 1.84𝜆/𝐵𝑎𝑐𝑜𝑠𝜃𝑎

Where 𝐿𝑐, and 𝐿𝑎 are the stacking height and the lateral size of the crystallite, respectively; 𝜆

is the wavelength of the used X-ray; 𝐵𝑐 and 𝐵𝑎 are respectively the full width half maximum

(FWHM) of the (002) and (100) peaks from the diffractogram of each sample, whereas 𝜃𝑐 and

𝜃𝑎 are the corresponding scattering angles.

The interlayer spacing 𝑑002 was determined using Bragg’s law: Eq 2.6.3

Eq 2.6.3 𝑑002 = 𝜆/2𝑠𝑖𝑛𝜃002

The number of crystallite layers per stack (N) was therefore calculated as: Eq 2.6.4

Eq 2.6.4 𝑁 = 𝐿𝑐/𝑑002 + 1

2.6.8 Mechanical characterization

A dynamic mechanical analyzer DMA (TA RSA3) equipped with a three-point bending

fixture (Lxlxh:10x12.5x19 mm) was used for mechanical characterization of MFC/LS/CP

carbon precursor and the resulting carbons. The tested samples were 3D printed mono-lines

that were (or not) subsequently carbonized to HTT from 400 to 1200°C. The standard of ISO

178 was followed except for the special dimensions of the samples. The cross section of each

samples is approximately an ellipse. As a result, the corresponding section modulus was used

to calculate the flexural stress and strain as well as elastic modulus. In order to estimate the

uncertainty caused by dimension measurement and inhomogeneity, 6 different positions were

measured in width and thickness by a Vernier caliper. The crosshead speed was set to 0.001

mm/s in order to obtain enough data before rupture. 3 samples per HTT were tested.

58

3. Results and discussion

59

3 Experimental results

3.1 Use of MFC/LS blends as carbon precursors: impact of hydrogel

rheology on 3D printing

3.1.1 Introduction

The first study conducted during this thesis focused on the rheological characterization of

MFC/LS hydrogels of different concentrations. A successful use of 3D printing for

manufacturing the carbon precursors requires a deep investigation of the used “ink”. For such

purpose, LS and MFC suspensions were examined in separate and mixed form. Both flow

mode and thixotropic mode were applied in order to test the usability of MFC/LS hydrogels in

different printing conditions. The relationship between the rheological behavior of MFC/LS

hydrogels and the 3D printing were concluded and evidenced by the photographs of the

printed objects. Drying conditions and some basic physical properties of the printed-

carbonized objects were also given with comparison to those from literature.

3.1.2 Pure LS/water solutions of different concentrations

Fig 3.1.1 shows that, regardless of concentrations, LS solutions displayed a Newtonian

behavior with a progressive viscosity increase from 0.003 to 100 Pa∙s when LS concentration

rose from 20 to 55%. The rheological behavior of LS solutions was therefore described by

plotting relative viscosity as a function of the volume fraction of dissolved LS (inset in Fig

3.1.1). Below a volume fraction of ca. 0.35, relative viscosities slightly increased with

concentration, whereas, above that value an abrupt increase was observed and for LS volume

fractions higher than 0.47 it was no longer possible to measure viscosity.

60

Fig 3.1.1 Rheological behaviours of LS solutions of different concentrations. The inset

represents the relationship between relative viscosities and volume fractions. Full points and

the dashed line represent experimental data and the fitting curve, respectively.

The Krieger-Dougherty equation(Eq 3.1.1)94,95

was used to fit experimental data and to

correlate the relative viscosity of the polymer solution (𝜂𝑟) to its volume dissolved fraction

(𝜙):

Eq 3.1.1 𝜂𝑟 = (1 −𝜙

𝜙𝑚)

−[𝜂]𝜙𝑚

where 𝜙𝑚 and [𝜂] are the maximum packing fraction (which corresponds to the critical phase

volume at which the fluid’s viscosity tends to infinity) and the intrinsic viscosity of LS,

respectively.

Values obtained for 𝜙𝑚 and [𝜂], i.e. 0.49 and 8.14, were in line with those obtained by Vainio

et al.94

for similar systems, thus indicating that intermolecular interactions above a volume

fraction of 0.49 lead to a fluid to solid transition and LS solutions are no longer processable

by extrusion. In order to limit the viscosity of MFC/LS suspensions the maximum LS volume

fraction used in this study was 0.42, corresponding to a LS mass fraction of 50% and a

solution viscosity of 3.3 Pa∙s.

3.1.3 Pure MFC hydrogels of different concentrations

Fig 3.1.2 shows that MFC hydrogels displayed a typical shear thinning behaviour65,66

, which

is associated to the progressive destruction of the MFC network upon shearing and was

modelled using the Herschel-Bulkley equations:

Eq 3.1.2 𝜎 = 𝜎𝑦 + 𝐾��𝑛 + 𝜂𝑠��

Eq 3.1.3 𝜂 = 𝜎𝑦

��+ 𝑛𝐾��𝑛−1 + 𝜂𝑠

where 𝜎 is the shear stress, 𝜎𝑦 the yield stress, �� the applied shear rate, 𝜂𝑠 the viscosity of the

Newtonian suspending medium (i.e. water or LS solutions), 𝜂 the suspension viscosity, 𝐾 the

concentration factor and 𝑛 the flow index which represents the fluid flow behavior (i.e. 𝑛 = 1

Newtonian, 𝑛 > 1 shear-thickening, 𝑛 < 1 shear-thinning).

61

Fig 3.1.2 Viscosity (a) and shear stress (b) responses of MFC suspensions at 4 different

concentrations in a shear flow. Dashed lines represent experimental data fitting with Eq 3.1.2

and Eq 3.1.3.

A rise in MFC concentration led to a general increase of the shear stress and viscosity, which

was mainly due to the progressive formation of dense nanofiber networks opposing a growing

resistance to shear67

. The plot of the yield stress and the flow index values obtained from Eq

3.1.2 as functions of MFC concentrations (Fig 3.1.3), shows that the dilution to 0.5% of the

pristine 2% MFC suspension led to the hydrogel disruption and a shift from shear thinning

towards Newtonian behaviour (i.e. the yield stress dropped from 4 to 0.3 Pa, and flow index

increased from 0.5 to 0.85).

Fig 3.1.3 Yield stress and flow index of MFC suspensions obtained from experimental data

fitting with Eq 3.1.2

Despite a high viscosity at low shear rate, the pronounced shear thinning and high yield stress

of 2 and 11.4% MFC suspensions led to hydrogel fluidization in the printing needle (for a

shear rate of ca. 300 s-1

, 2% and 11.4% hydrogels dropped to ca. 0.1 and 2 Pa∙s, respectively).

As highlighted in previous works171

, 3D objects can be easily printed. Whereas, 0.5% MFC

suspensions behaved as low viscosity Newtonian fluids and spread over the printing substrate.

62

3.1.4 MFC/LS slurries of different concentrations

Flow curves for mixed MFC/LS suspensions show that, at 2% MFC (Fig 3.1.4a)), the increase

of the LS concentration results in a general shift of the flow curve towards higher viscosity.

This is in line with the viscosity increase in the suspending medium.

Whereas, at 1 and 0.5% MFC (Fig 3.1.4b) &c)), LS addition induced a drop in viscosity at

shear rates below ca. 100 and 10 s-1

, respectively. Above those shear rate values, the

viscosities of MFC/LS systems were higher than those of pure MFC gels indicating that at

low shear rates the viscosity of MFC/LS mixtures is dictated by the disruption of weak MFC

networks. At high shear rates, the rheology is dominated by the viscosity of the suspending

medium. Regardless of MFC concentrations, in the presence of 50% LS, the viscosity of

MFC/LS systems was higher than that of the pristine MFC hydrogel and at high shear

converged to 𝜂𝑠.

The plots of the yield stress values obtained from fits of Eq 3.1.2 to the experimental data (Fig

3.1.5) show that at concentrations below 30%, LS induced a drop in the MFC gel strength.

Indeed, a drop in yield stress was clearly detected with 1 and 0.5% MFC. At a MFC

concentration of 2%, only a slight decrease was observed.

As highlighted for diluted MFC/soluble polysaccharides mixtures172

, the sharp decrease in the

yield stress at extremely low LS concentration (i.e., 0.5% LS) is associated with the

adsorption of LS on the fiber surface; the adsorbed LS, acting as lubricant, decreases the

friction among nanofibers and the stress necessary to break the entangled nanofiber network.

Increasing the LS concentration from 0.5% to 30% poorly affected the yield stress, thus

indicating that over this LS concentration range the LS no longer affects the interfiber friction.

Therefore, we can go on the assumption that between 0.5% and 30% LS, the fiber surface is

saturated by adsorbed LS molecules and their effect on interfiber lubrication attains a steady

state. Moreover, dissolved LS molecules have limited or no effect on fiber interactions. The

minor effect of LS on the yield stress at an MFC concentration of 2% is ascribed to the

formation of dense networks of entangled fibers, where interfiber friction is supposed to make

a minor contribution to the yield stress compared with the mechanical fiber bending and

deformation necessary to break the MFC network.

Above a LS concentration of ca. 30-40%, corresponding to an abrupt increase in the viscosity

of the LS solution (Fig 3.1.1), the yield stress increased for all tested MFC concentrations

attaining values close to that of pristine MFC hydrogels.

63

Fig 3.1.4 Plots of viscosity vs shear stress for MFC/LS systems: (a) 2% MFC series (the inset

shows shear stress vs shear rate data); (b) 1% MFC series; (c) 0.5% MFC series. Dashed lines

represent fits of Eq 3.1.2 and Eq 3.1.3 to the experimental data.

Fig 3.1.5 Influence of the MFC mass fraction (colored symbols) and the LS mass fraction on

the yield stress. The yield stress values were obtained from fits of Eq 3.1.2 to the experimental

data.

64

At high LS concentration, interfiber lubrication by LS plays a secondary role and the strength

of the MFC network is supposed to be mainly affected by the high viscosity of the suspending

medium, which, by decreasing fiber mobility, increased the stress necessary to break the MFC

network (i.e. the yield stress).

Overall, at concentrations below 30% the viscosity of LS solutions is below 10 Pa∙s, and LS

lubricates nanofiber contact nodes and favors their relative motion and network disruption

when shear is applied. At higher concentration, the high viscosity of LS solutions (𝜂 > 70

Pa∙s) screens the lubricating action by inducing a drop in fiber mobility thus restoring the

resistance of the network to shear.

3.1.5 Thixotropy of mixed slurries

Under oscillatory deformation, MFC suspensions are characterized by viscoelastic response

with high storage modulus and low phase shift172,173

. This quasi-elastic behavior indicates that

structural changes in the MFC network (i.e., breakdown and rebuild) induced by variations in

the shear rate occur with no or short time delay. Nevertheless, Sorvari et al.172

showed that

carboxymethylcellulose (at a concentration of 0.11%) increased the viscosity of the

suspending solution up to ∼80 Pa∙s and the viscous behavior of the MFC hydrogel,

corresponding to an increase in the time delay for structure change in the MFC network.

In order to investigate the role of LS in affecting the MFC network rebuild kinetics, MFC/LS

suspensions were subjected to a step-down shear variation simulating the shear conditions

applied during the suspension extrusion through the 3D printer needle. Network rebuild after

shearing was described using a first-order kinetic model to describe the response of the MFC/

LS suspension71

:

Eq 3.1.4 𝜎(𝑡) = 𝜎0 + (𝜎∞ − 𝜎0) (1 − 𝑒−𝑡

𝜏) + 𝜂𝑠��

where 𝜎0 is the shear stress immediately after the step-down variation in the shear rate, 𝜎∞ is

the shear stress after an infinite time, 𝜏 is the time constant, and 𝜂𝑠 is the viscosity of the

Newtonian surrounding medium (i.e., the bare LS solution). Fig 3.1.6 shows that in MFC/LS

systems with LS concentrations below 30%, the shear stress exhibited a quick monotonic

decrease, reflecting the viscoelastic response of the MFC network to the sudden shear

variation174

. The time constant obtained from fits of Eq 3.1.4 to the data progressively

increased from 4 to 7 s as the LS concentration increased from 0% to 30%, thus indicating

that LS slowed the response of the MFC network.

65

Fig 3.1.6 Shear stress responses to the step-down experiments for the 2% MFC series. The

shear rate dropped from 1000 to 0.1 s−1. Dashed lines represent fits of Eq 3.1.4 to the data.

The shear stress drop and subsequent increase until stabilization observed at an LS

concentration of 40% revealed a transition from viscoelastic behavior to inelastic thixotropic

behavior. As a result of the increased viscosity of the LS solution, rebuilding of the MFC

network was delayed and, as observed for papermaking fibers175

, occurred progressively after

the step-down shear variation. Increases in the LS concentration up to 45% and 50% further

emphasized the thixotropic response, with increases in the time constant from 0.7 s at 40% LS

to 14 and 27 s at 45% and 50% LS, respectively.

3.1.6 Relationship between the rheology and hydrogel printing

As illustrated in Fig 3.1.7, square cuboids with good spatial definition (i.e., sharp edges and

no spreading of the cuboid base on the substrate) were printed using the bare (2%) MFC

hydrogel. Additions of LS led to a progressive degradation of cuboid shape:

i. With up to 30% LS, the cuboid base was subjected to ever-increasing spreading, and

the height of the printed object decreased from about 9 to 2 mm as the LS

concentration was increased from 0 to 30%. According to the experimental data shown

in Fig 3.1.6, this behavior was associated with a decrease in the viscosity (from 130 to

60 Pa∙s) at low shear rates.

ii. With 40% LS, the MFC/LS suspension completely spread over the printing substrate,

and the thickness of the final liquid film was below 1 mm. This trend was ascribed to

the transition from viscoelastic to inelastic thixotropic behavior. Indeed, after shearing

in the extruder needle and destruction of the MFC network, the flock rebuild kinetics

was too slow and the fluid viscosity too low to stop the MFC/LS suspension from

spreading over the printing substrate.

iii. With 50% LS, square cuboids were easily printed without major deformations.

However, their edges displayed rounded profiles, indicating that the flow of the

MFC/LS suspension did not stop immediately after extrusion. This behavior was

66

associated with the high viscosity of the suspending medium (η = 3.3 Pa∙s), which,

despite the pronounced thixotropic behavior and long MFC network rebuild time (τ =

27 s), slowed the spreading and provided sufficient time for network rebuild and

object geometry “freezing” before major deformations were induced.

As highlighted in Fig 3.1.7, only the samples with LS concentrations of 0−10% and 50% held

the pristine geometry after printing. However, with 0−10% LS, freeze-drying was necessary

to hold the original shape since air drying induced extensive shrinkage. With 50% LS, air

drying led to a limited deformation of the square faces and 40% shrinkage along the vertical

axis, which for the purpose of this study was considered acceptable.

Fig 3.1.7 Shape and main characteristics of 3D printed cuboids before and after drying and

carbonization. The red line represents a length of 10 mm.

3.1.7 Characterization of printed aerogels before and after pyrolysis

X-ray tomography of the cuboid sections (Fig 3.1.8) showed that before and after

carbonization, the freeze-dried sample containing 4% LS (corresponding to a dry sample

composition of 33.3% MFC and 66.7% LS) displayed a lamellar structure associated with

solute accumulation at the boundaries between ice crystals, which acted as template176,177

.

According to the slow convective freezing used in this study, Fig 3.1.8 shows fine and

continuous lamellar structures originating from the cuboid surface, converging to the cuboid

core, and forming a continuous network. After carbonization, this particular structure led to

the formation of an extremely light carbon cuboid with an apparent density of ca. 63 kg m−3

and an electronic conductivity of ca. 5.5−55 S m−1

.

LS: 0% 10% 30% 50%

Before drying

After air drying

Freeze drying necessary to hold the original shape

Not adapted for 3D printing

Lateral side Lateral sideTop side Top side

After freeze drying and pyrolysis (LS 4%)

After air drying and pyrolysis

LxWxH: 1.4x1.4x0.5 ± 1 mmMass: 0.061 grapp: 63 ± 20 kg m-3

sel: 5.5 – 55 S m-1

LxWxH: 1.8x1.8x0.8 ± 1 mmMass: 1.11 grapp: 428 ± 100 kg m-3

sel: 3 – 21 S m-1

67

Fig 3.1.8 X-ray tomography sections of freeze-dried cuboids from 2% MFC/4% LS

suspension (i.e., 33.3% MFC and 66.7% LS in the dry sample): a) and b) before pyrolysis; c)

and d) after pyrolysis

Fig 3.1.9 X-ray tomography sections of air dried cuboids from the 2% MFC/50% LS

suspension (i.e., 3.8% MFC and 96.2% LS in the dry sample).

For the sample containing 50% LS, the lamellar structure was absent after freeze-drying, and

the solid phase displayed the presence of fine cracks, presumably due to the formation of

small ice crystals. As highlighted for inorganic hydroxyapatite particles177

, this morphology

change was interpreted as indicating that LS was not repelled from the ice front during crystal

growth.

68

The air-dried sample with 50% LS (corresponding to a dry sample composition of 3.8% MFC

and 96.2% LS) had high internal porosity with macropores and large cracks (Fig 3.1.9). This

irregular structure was associated with the presence of large amounts of residual water in the

sample before carbonization. Indeed, as determined by direct weighing, the water content

decreased from 64% to 38% and 30% (w/w) after drying in air and at 110 °C, respectively.

Despite the high density of the continuous phase (highlighted by the pronounced contrast in

the X-ray tomography images) and the high apparent density of the carbonized cuboid (i.e.,

428 kg m−3

), the electronic conductivity along the Z axis was lower than that of the freeze-

dried sample, (i.e., 3−21 S m−1

). This difference is associated with discontinuities in the

conductive path generated by macropores and cracks. This work shows that microfibrillated

cellulose/lignosulfonate hydrogels appear to be promising bio-sourced precursors for the

manufacture of carbon objects with simple geometries by 3D printing and carbonization. Even

if several aspects concerning the control of object shaping by 3D printing and its geometric

stability during drying/carbonization remain to be explored, carbon cuboids obtained by

freeze-drying from a dry precursor made of 33% MFC and 67% LS displayed interesting

properties (i.e., low density, geometric stability, and homogeneous structure). Moreover,

compared with carbon materials obtained from other wood-derived precursors (see Table

3.1.1), they display extremely low density and good electric conductivity, making them

promising materials for the manufacture of 3D-structured porous electrodes for energy storage

devices.

Table 3.1.1 Comparison of electrical conductivity of wood-derived carbon materials.

Reference

work

Conductivity

(S m-1)

Density

(kg m-3) Physical form Precursor

This work 5.5-55 63 ± 20 3D printed cuboid MFC/Lignosulfonate blend

Teng et al.178

2.3-3 nd Fibre, 639-816 nm diameter Kraft Lignin/MW Carbon

Nanotube blend

Snowdon et

al.179

0.9 nd Powder compressed at 1.12 MPa Hydrolysis Lignin

Deraman et

al.180

243-500 1000 Bulky pellet Lignin/precarbonized carbon

Rhim et al.10

10000 2000 Bulky pellet Microcrystalline cellulose

*Lu et al.181

* 90-2000 65-600 Monolithic aerogel Polycondensed resorcinol-

formaldehyde gel

*Sanchez-

Gonzalez et

al.182

*

44-215 250-700 Compressed carbon black Carbon black

*Carbon from sources other than wood given for comparison

3.1.8 Conclusions

MFC/LS systems displayed a complex rheological behavior that was affected by the LS

concentration. At low concentrations (i.e., below 30%), LS acted as a lubricant, favoring

shear-induced disruption of the MFC network and inducing a progressive decrease in both the

yield stress and viscosity. The print quality of the cuboids underwent progressive degradation

69

due to spreading of the MFC/LS over the printing surface, and 10% was the maximum LS

concentration that could be used to print cuboids.

At intermediate LS concentrations (ca. 40%), the MFC/LS suspensions shifted from

viscoelastic to inelastic thixotropic fluids with complete degradation of the print quality, i.e.,

the MFS/LS suspension completely spread over the printing substrate. At high LS

concentrations, the print quality the cuboids was restored since despite the pronounced

thixotropic behavior, the high viscosity of the suspending medium slowed the fluid spreading

during printing, thus providing sufficient time to rebuild a continuous MFC network and

“freeze” the cuboid geometry.

Freeze-drying and carbonization of cuboids printed using MFC/LS suspensions with a

maximum LS concentration of 10% yielded carbon cuboids with minimal shape variations

that displayed an ice-templated continuous lamellar structure, extremely low apparent density

(ρapp = 63 ± 20 kg m−3), and high electrical conductivity (σ = 5.5−55 S m−1). Because of the

high dry material content, cuboids printed with 50% LS suspensions underwent acceptable

shrinkage upon air drying and carbonization. Nevertheless, despite interesting conducting

properties, large defects (i.e., macropores and crackles) generated during drying led to a

heterogeneous and brittle object.

Overall, this study demonstrates that MFC/LS hydrogels have the potential to be used as

biosourced precursors for the manufacture of 3D-printed carbon objects, but work remains to

be done to demonstrate the level of control that would be necessary for a manufacturing

environment.

70

3.2 Use of MFC/LS blends as carbon precursors: pyrolytic process

characterization and kinetics study

3.2.1 Introduction

At the end of the chapter 3.1, conductive and self-standing objects were successfully

elaborated using 3D printing and the subsequent pyrolysis. Since pyrolysis is the fundamental

process that transforms carbon precursors into engineering carbons and plays a significant

role in determining carbons’ properties (morphology, electrical conductivity, etc.), a further

study of the thermal degradation of MFC/LS composites and their kinetics is quite necessary.

This second part of chapter 3 presents the thermal characterization of MFC/LS blends using

TGA/MS. The kinetic analysis was conducted using both model-free and mode-based

methods. Arrhenius parameters were given for MFC and LS separately as well as in

composites.

3.2.2 Morphological and structural characterization

MFC and NaLS alone, together with one of their composite are observed under SEM, as

shown in Fig 3.2.1. MFC nano-paper, made by water evaporation from the original 2%

aerogel, is in the form of a thin film in macroscale, whereas in microscale, it consists of a

stack of micro-layers in the thickness direction (Fig 3.2.1b))183

. Each micro-layer is formed by

cellulose microfibrils entangling with each other and exhibits a web-like structure184,185

, as

observed in Fig 3.2.1a). Dimensions of single microfibril of cellulose are beyond the maximal

resolution of the SEM to be distinguished. Dissimilarly, industrial NaLS is in powder-form

and SEM image shows that these powders are essentially spherical but hollow (indicated by

arrows in Fig 3.2.1c) with diameters ranging between 20 and 135 μm90

. Composite of NaLS

and MFC (66% and 33%, respectively) is examined on fraction surface and as shown in Fig

3.2.1d). A sheet-like structure of MFC can be observed as in pure MFC film, whereas NaLS,

blended previously with MFC in suspension state, acts like “glue” which surrounds MFC

sheets spatially and sticks them together in transversal direction to form a compact stack after

water evaporation.

Fig 3.2.2a) illustrates chemical structures patterns of single MFC, NaLS along with their

composites using FTIR. Characteristic bonds of MFC and NaLS are identified directly in Fig

3.2.2b), with good similarity to those reported in the literature186–188

. MFC/NaLS blends do

not present other chemical bonds than those inherited from single MFC and NaLS. Besides,

the specific bands related to each constituent are more or less pronounced depending on the

fraction of MFC and LS in the composite.

71

Fig 3.2.1 SEM micrographs of: (a) MFC film surface; (b) MFC film in transversal sense: (c)

NaLS powder (arrows indicate the hollow nature of NaLS); (d) NaLS/MFC composite

(66%/33%) in fraction surface

72

Fig 3.2.2 FTIR spectra of MFC/NaLS composites of various ratios: (a) global comparison; (b)

characterization of chemical bonds in MFC and NaLS as a function of wavenumber

3.2.3 Thermal characterization

Thermal degradation patterns of MFC and NaLS as well as their different composites are

presented in Fig 3.2.3. One can see on this figure that the thermal degradation characteristics

of MFC and NaLS are quite different. Unsurprisingly, MFC presents a very similar thermal

decomposition pattern to that of cellulose as reported in the literature122–125

since they both

have the same chemical structure owing to their common cellulose nature. Pure cellulose

decomposes mainly from 250°C to 400°C and exhibits a sharp reaction peak at around 334°C.

In the end of pyrolysis, the carbonaceous char is formed and weights slightly more than 20%

of the initial mass of the sample. Compared to MFC, DTG curve of NaLS is more

complicated and is composed of two successive reaction peaks (at 239°C and 291°C) and four

visible shoulders (at around 200, 330, 410 and 665°C, respectively), not far from previous

reports concerning LS degradation129,131,189

. NaLS thermal degradation starts around 150°C.

Its thermal degradation process takes place in a large temperature range up to 800°C.

Nevertheless, NaLS degrades more mildly than MFC and thus results higher char yield at

around 47%90

.

Composites of MFC and NaLS seem to combine the thermal behavior of both pristine

materials. They appear also to exhibit more similar thermal degradation patterns to the

predominant constituent in them. For instance, compared to pure MFC, adding 23% of NaLS

into MFC makes such a composite to degrade more prematurely with a shift of maximum

reaction rate towards lower temperatures by 26°C whereas the morphology of its DTG curve

still recalls the characters of cellulose decomposition. Likewise, when the content of NaLS in

the composite becomes predominant, which is the case for the composite of 16% MFC and 84%

NaLS, it follows roughly the degradation pathway of single NaLS. The comparatively huge

73

peak at around 250°C that this composite shows is most probably the superposition of one

reaction peak of NaLS originally at 239°C, and the characteristic peak of MFC decomposition

initially at 334°C in pure MFC pyrolysis. This peak shifts towards lower temperature regions

with increasing NaLS content in the composite. Char mass increases with increasing NaLS

content in the composites, except for the composite with 84% NaLS. The latter gives a char

yield slightly higher than in pure NaLS.

Fig 3.2.3 (a) TG and (b) DTG curves of MFC, NaLS (shorted to LS) and their composites (the

arrow indicates the shift of characteristic peak of cellulose)

3.2.4 Catalytic effect of sodium

For composites where the MFC content is predominant (MFC: 77%, 67% and 50%), the shift

of characteristic peak for cellulose decomposition, that is indicated in Fig 3.2.3b) by an arrow,

is thought to be due to the catalytic effect of sodium ions that NaLS brings in129,189

. Although

NaLS degrades earlier and slower than MFC, there are no reasons - except interactions

between NaLS and MFC- to shift the degradation peaks at low temperatures. In the absence of

interactions, the composites should degrade following a TG/DTG curve which would

correspond to the linear combination of that of the pure MFC and NaLS. The theoretically

predicted TG/DTG curves are calculated according to Eq 3.2.1 or Eq 3.2.2 based on the

experimental TG/DTG data of MFC and LS obtained separately. As shown in Fig 3.2.4 a) and

b), obvious differences between the “theoretical” curves and the experimental ones are

obtained for both MFC-predominant composites. When the NaLS content rises from 23% to

50%, the sodium content in the composite increases from 2% to 4.5%, making cellulose in

these composites to decompose earlier and earlier due to an increasing catalytic effect. The

peak temperatures shift to 308°C, 296°C, and 283°C, respectively for composites with 77%,

67% and 50% of MFC, in comparison with the peak temperature observed at 334°C for pure

MFC. The discrepancies between the theoretical additive curves and the experimental ones

are more pronounced with increasing sodium content. Reasonable explanation should still

behind the catalytic effect of Na that makes cellulose characteristic peak to get closer to that

74

of NaLS. Therefore, because of the “superposition effect”, the characteristic region of NaLS

is reinforced and appears more obviously in DTG curves of composites.

The temperature shift ∆Tp (representing the difference between the observed peak temperature

in the composite and that observed in pure MFC) is plotted as a function of the Na content in

the composite as shown in Fig 3.2.5a). One can observe a clear linear dependence between

these two variables, which strongly suggests that the presence of Na is the main reason for the

peak shift in the DTG curves.

Furthermore, we investigated the effect of NaCl impregnation on MFC on its thermal

behavior. The results are plotted in Fig 3.2.5b) and suggest that sodium ions, introduced

otherwise into MFC, are indeed capable of decreasing MFC reaction rate as well as shifting

the degradation peak to lower temperature regions. However, compared to MFC/NaLS blends,

the different mobility of Na+ together with the presence of chloride ions in MFC/NaCl

mixture catalyze MFC pyrolysis in a slightly different way by yielding a smaller ∆Tp despite

the higher Na content.

Eq 3.2.1 𝑇𝐺(𝑇) 𝑇ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙 𝑎𝑑𝑑𝑖𝑡𝑖𝑣𝑖𝑡𝑦 = % 𝑀𝐹𝐶 × 𝑇𝐺(𝑇)𝑀𝐹𝐶 + %𝐿𝑆 × 𝑇𝐺(𝑇)𝐿𝑆

Eq 3.2.2 𝐷𝑇𝐺(𝑇) 𝑇ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙 𝑎𝑑𝑑𝑖𝑡𝑖𝑣𝑖𝑡𝑦 = % 𝑀𝐹𝐶 × 𝐷𝑇𝐺(𝑇)𝑀𝐹𝐶 + %𝐿𝑆 × 𝐷𝑇𝐺(𝑇)𝐿𝑆

Fig 3.2.4 Comparison of theoretical and experimental TG/DTG curves for two MFC-

predominant composites: (a) MFC77% LS23%; (b) MFC50% LS50%

75

Fig 3.2.5 a) Influence of sodium content on the observed ∆Tp in the MFC/NaLS blends; b)

Evidencing the catalytic effect of Na by the TG analysis of MFC/NaCl blend

In the study of Wang et al.190

, it was observed that not only NaCl, but also other sodium

compounds, such as Na2CO3, NaOH and Na2SiO3, could make wood decomposition to take

place under lower temperatures. Sodium ion can penetrate into the biomass textures and break

the intermolecular hydrogen bridges under swelling or heating190,191

. As a result,

devolatilization occurs earlier.

3.2.5 Pyrolysis gas analysis by mass spectrometry (MS)

Fig 3.2.6 illustrates the intensity evolution of the major pyrolysis products throughout the

high temperature pyrolysis of the different MFC/LS composites. The TG/MS experiments

were performed without a calibration of the mass spectrometer. The intensities obtained for

the selected molecules were blank corrected considering the signals obtained for pure Helium

gas, and normalized to the initial sample mass. The results presented in Fig 3.2.6 can only be

compared qualitatively192

. The intensity evolution of H2O molecule roughly follows the

features of the DTG curve regardless of the sample , which is in agreement with the fact that

water is the most important volatile product131,189

. It is believed that the cracking of alphatic

hydroxyl groups (-OH band at 3350cm-1

in Fig 3.2.2b) for both MFC and LS) in the lateral

chains generates water131

. Another strongly emitted gaz is carbon dioxide. According to MS

plot of CO2, the decarboxylation reaction is apparently composed of two stages: the first one

begins quite early around 150°C and lasts continuously till 500°C; the second stage takes

place between 500 and 700°C and is particularly remarkable for pure NaLS sample. The

complicated evolution of CO2 emission should be attributed to the various functional groups

existing in MFC and NaLS, including carboxyl, carboxylate, ester, carbonate groups among

others. The specially strengthened decarboxylation in NaLS at around 640°C is possible to be

a result of the decomposition of Na2CO3, an intermediate matter produced during pyrolysis129

.

Distinguished from CO2, CO is released quite late in the temperature range of 650 to 750°C

for NaLS, in contrast with MFC, for which the CO intensity reached its maximum at around

76

325°C. Surprisingly, composites (LS23% and LS84%) do not present an obvious emission

maximum at around 700°C as LS does, which seems that MFC inhibits the CO production

during their high-temperature devolatilization stage. Moreover, a large amount of alkyls and

aromatic alkenes with various molecular masses are released during intermediate temperature

range. The most important product, methane (m/z=15), whose emission evolution gives a clue

concerning the demethylation of biomass that consists of two successive reaction peaks at

approximately 300 and 500°C regardless of the sample. Decomposition of sulfonates, where

the latter are represented by S-O band (650 cm-1

), C-S band (1028 cm-1

) and S=O band

(1200cm-1

) in FTIR results in Fig 3.2.2b), is examined according to H2S and SO2

emissions193,194

. SO2 is strongly released at early stage from 200 to 350°C in NaLS, signifying

that the sulfonic acid group is rather thermally instable. It is interesting to observe that

although H2S is not largely present during pure NaLS pyrolysis, whereas it is strongly

released in the composite MFC16%/LS84%. It is reasonable to assume that MFC could

strengthen the H2S production from NaLS but the true mechanism has to be further

investigated.

Fig 3.2.6 MS plots of major pyrolysis gases during the TG experiments of the different

MFC/LS composites

3.2.6 Kinetic analysis

3.2.6.1 Estimation of activation energies by KAS isoconversional method

KAS isoconversional method is employed to estimate the activation energies at the different

conversion level (0.1-0.9) during the pyrolysis of MFC and NaLS up to temperature of 500°C.

Arrhenius plots for both materials based on four heating rates are presented in Fig 3.2.7.

77

Highly linear relationship between ln(𝛽𝑗/𝑇𝛼,𝑗2 ) versus 1/Tj confirms the hypothesis that

activation energy should be constant at fixed conversion regardless of heating rates.

Activation energies for each conversion level are resumed in table.1.

Fig 3.2.7 Arrhenius plot for (a) MFC and (b) LS at conversion levels from 0.1 to 0.9

Table 3.2.1 Activation energies provided by KAS method

α MFC NaLS

Ea (kJ/mol) R2 Ea (kJ/mol) R

2

0.1 206.62 0.994 197.49 0.996

0.2 223.37 0.984 243.77 1.000

0.3 237.9 0.962 264.82 0.999

0.4 240.45 0.952 299.26 0.994

0.5 244.86 0.947 246.57 0.996

0.6 248.02 0.945 235.91 1.000

0.7 253.69 0.957 246.8 0.999

0.8 268.77 0.969 307.07 0.996

0.9 763.78 0.959 542.19 0.984

It can be seen from Table 3.2.1 Activation energies provided by KAS method that E varies at

different conversion levels for both MFC and NaLS. At most of conversion levels, MFC has

an activation energy that varies between 206 and 270 KJ/mol and rises with increasing

conversion content. E of NaLS increases sharply at first (from 197 to 300 KJ/mol, when α

rises from 0.1 to 0.4), and then followed by a slight decrease back to 246 KJ/mol for α from

0.5 to 0.7. E values for MFC thermal degradation are quite comparable with those provided in

the literature and will be detailed in the following section. However, to the authors’ best

knowledge, no kinetic data for NaLS thermal degradation are published. For both MFC and

78

NaLS, an abnormal divergence of E could be observed at α>0.8, which demands further

investigation.

3.2.6.2 Kinetic triplets provided by model-fitting method

The number of pseudo-components required for modeling was rather difficult to fix a priori.

Especially for the LS sample, which exhibited a quite complicated DTG curve pattern,

indicating the occurrence of several reactions taking place in different temperature ranges.

The choice was made leaning on the DTG curve shape (peaks, shoulders) as well as on the

MS spectra. According to conversion features of MFC and NaLS, they are deconvoluted into

4 and 6 pseudo-components, respectively.

Fig 3.2.8 Kinetic modelling of MFC: (a) TG modelling; (b) DTG modelling; (c) relative

deviation

Fig 3.2.9 Kinetic modelling of NaLS: (a) TG modelling; (b) DTG modelling; (c) relative

deviation

79

The confrontation between the experimental data and the models are shown in Fig 3.2.8 and

Fig 3.2.9 respectively for MFC and NaLS. As it can be seen on these figures, the proposed

models reproduce quite well the TG data. The relative deviation as a function of temperature

is calculated according to Eq 2.5.7 and is presented in Fig 3.2.8c) and Fig 3.2.9c) whose

maximum is less than 3% for both materials. The identified proportion, activation energy and

pre-exponential factor of each pseudo-component are listed in Table 3.2.2 and Table 3.2.3.

The last decomposition reaction for MFC may be related to decarboxylation and

decarbonylation reactions of the char since CO2 and CO are the only two volatiles produced in

this range of temperature according to MS plots. With regard to the LS sample, the last

reaction is more probable to be the thermal decomposition of an intermediate product during

pyrolysis -Na2CO3189

since there is a clear emission peak of CO2 suggested by MS spectra in

the same temperature range. However, even with the help of mass spectrometry, it is hard to

identify the nature of other pseudo-components since they are normally the sum of several

reactions.

Table 3.2.2 Arrhenius parameters for the 4 pseudo-components of MFC

Pseudo-components 1 2 3 4

Proportion 0.22 0.51 0.18 0.09

Ea (KJ/mol) 137 229 46 52

logA (log(s-1

)) 23.7 40.3 1.5 0.1

Table 3.2.3 Arrhenius parameters for 6 pseudo-components of NaLS

Pseudo-components 1 2 3 4 5 6

Proportion 0.05 0.17 0.19 0.15 0.24 0.20

Ea (KJ/mol) 149 112 132 104 45 72

LogA (log(s-1

)) 33 21 23 15 1.2 2

3.2.6.3 Comparison between isoconversional and model-fitting methods

The Ea values provided by isoconversional approach (Table 3.2.1) are generally higher than

those obtained by the model-fitting approach for both materials regardless of conversion

levels. Particularly, for conversion levels higher than 0.8, a clear divergence of E values could

be perceived for both MFC and NaLS when comparing the two approaches. Furthermore,

using even the smallest E value from table.1 to model the thermal degradation of pseudo-

component 3 or 4 of MFC could not result a fitting as good as the original value does, despite

the attempt to adjust the pre-exponential factor value to recompense the peak form change.

These observations also hold for NaLS fitting procedure using values from Table 3.2.1.

Due to lack of kinetic data for NaLS pyrolysis in the literature, only MFC kinetic parameters

are discussed by comparing them with those of cellulose from already published articles. As

shown in Fig 3.2.10. E values suggested from literature as well as from this work suffer a

great discrepancy depending on biomass types, experimental conditions, the choice of kinetic

analysis approaches, etc.. The results obtained from isoconversional method are usually

higher than those provided by the model-fitting approach, especially for high temperature

80

range reactions (e.g. pseudo-component 3), which is in agreement with our observation above.

Furthermore, the activation energies given by model fitting in this work are quite close and

consistent with those reported in literature, which confirms as well the accuracy of such

approach being employed in kinetic analysis of MFC/NaLS.

Fig 3.2.10 Comparison of Ea values provided in this work and those from references in terms

of two methods: article 1195

; article 2196

; article 3127

;article 4107

3.2.6.4 Kinetic analysis of MFC/NaLS composites

Kinetics of such composites was analyzed based on single MFC and NaLS model.

Consequently, each composite should contain ten pseudo-components whose proportion is

determined by multiplying its original value in MFC (or NaLS) by the corresponding fraction

of MFC (or NaLS) in the composite. From a kinetic point of view, presence of sodium ions

catalyzes cellulose decomposition causing the activation energies of the latter to diminish in

composites102

. Therefore, lower activation energies should be expected for related pseudo-

components of MFC in composites, whereas for NaLS they remain unchanged. Besides, pre-

exponential factor A continues to be the same as in single MFC and NaLS kinetic model.

Under these conditions, model-based method still results good fitting for two MFC/NaLS

composites with relative deviation less than 3% throughout the whole temperature range,

justifying the assumptions of i) linear superposition of MFC and NaLS in the composites and

ii) drop of E due to catalytic effect. The utilized kinetic parameters are listed in Table 3.2.4.

Table 3.2.4 Kinetic triplets determined by model-fitting for 10 pseudo-components of

Composite MFC77%/LS23% and MFC16%/LS84%

81

MFC77%

MFC16% 1 2 3 4 5 6 7 8 9 10

Proportion

(10-2

)

16.94

3.52

39.27

8.16

13.86

2.88

6.93

1.44

1.15

4.20

3.91

14.28

4.37

15.96

3.45

12.60

5.52

20.16

4.60

16.80

Ea (KJ/mol) 132

126

219

197

45

43

52

50 149 112 132 104 45 72

LogA (log(s-1

)) 23.7 40.3 1.5 0.1 33.0 21.0 23.0 15.0 1.2 2.0

3.2.7 Conclusions

In this work, MFC and LS were characterized in both separate and blended form in terms of

morphology, chemical structure and thermal stability. Significant differences have been

observed between these two main materials in all aspects owing to their different chemical

constitution and structure. MFC degrades intensively within a narrow temperature range

between 250°C and 350°C, while NaLS decomposes much more gently within 2 main

temperature ranges, 150°C -500°C and 600°C -800°C. Moreover, kinetic analysis of separate

MFC and NaLS has been conducted with both model-free and model-based approaches. The

former method suggests Arrhenius parameters that are too high to be used to reproduce the

experimental curve. However, model-fitting approach results in a quite good reproducibility

of the experimental curve by providing MFC thermal degradation parameters that are quite

consistent with the bibliographic ones. Kinetic analysis of the thermal degradation of NaLS is

conducted for the first time providing a set of kinetic data that allows predicting the thermal

behavior of such material. Furthermore, DTG curves, that represent the decomposition of

MFC/NaLS composites mixed in various ratios, do not “linearly” combine the characteristic

patterns of both materials due to the presence of sodium content. The catalytic effect of

sodium is confirmed by a highly correlated relationship between Tp shift and sodium content

as well as experimentally by the resulting thermal behavior of a MFC/NaCl blend. From a

kinetic point of view, such a diminution of reaction temperatures could be interpreted as a

decrease of activation energy of MFC pseudo-components in composites. By using lower E

for MFC pseudo-components, successful fitting has been achieved for two MFC/NaLS

composites by following a weighted superposition rule.

82

3.3 Use of MFC/LS blends as carbon precursors: characterization of

MFC/LS carbons

3.3.1 Introduction

After rheological and thermal characterization, this part of work investigated the development

of electrical conductivity in carbonized carbons from MFC/LS precursors as function of HTTs.

MFC/LS carbons were characterized in terms of morphology (scanning electron microscopy),

chemical functionalities (infrared spectroscopy), microstructure (Raman spectroscopy and X-

ray diffraction) and physical properties (electrical conductivity and density evolution). A

descriptive model, based on the progressive conversion of the biomass into conductive

engineering carbons and composed of 3 distinguish phases, was thus established to illustrate

the electrical conductivity development phenomenon.

3.3.2 Morphology of MFC/LS sheet before and after pyrolysis

Fig 3.3.1 shows the pyrolysis-induced morphological changes of MFC16%/LS84% sheets in

the transversal section (a-c) and on the surface (d-f). The bulk structure of the composite is

preserved even after a pyrolysis at 1000°C197

, indicating a good morphological stability of the

MFC/LS system. This may be related to the slow heating rate adopted in the pyrolysis process

which preserves more the initial structure of the biomass than in the case of the fast pyrolysis.

Besides, the high percentage of LS which is quite heat-resistant129

should also be accounted

for such a good thermal stability. However, as shown in Fig 3.3.1b), a huge shrinkage in the

thickness direction was observed after carbonization at 400°C. This shrinkage was drastic at

the initial stages of pyrolysis and did not increase when increasing the HTT to 1000°C.

According to the thermogravimetric analysis 157

, the devolatilization peak of such composite

occurs at around 275°C and by 400°C, 80% of the total mass loss was achieved, which

explains the small morphological differences in chars carbonized at 400°C and at 1000°C.

The meso- and micro-porosity are invisible in the present scale of SEM, thus only macro-

porosity will be discussed herein. Fissures and pores that appear on the sectional images in

Fig 3.3.1a)-c) increases with the pyrolysis temperatures both in size and number. These

fissures probably originate from the volatile matter release and internal overpressure during

the pyrolysis process. It has to be noted also that some fissures were visible during ambient air

drying of the sample (as in Fig 3.3.1a)). However, the macropores density on the surface of

the chars (Fig 3.3.1e)&f) is not visibly much higher than non-carbonized one

83

Fig 3.3.1 SEM images of a)-c) the transversal section and d)-f) surface of MFC/LS sheet

before and after carbonization at 400 and 1000°C

3.3.3 Chemical characterization by FTIR

The FTIR spectra of MFC/LS chars prepared at 400-1200°C are displayed in Fig 3.3.2. It can

be observed that FTIR signals progressively lose their characteristics peaks with increasing

HTTs from 400°C. One of the most important reductions occurs in the wavenumber range of

3600-3200cm-1

, attributed to the hydrogen bonded O-H stretching vibration198,199

, which

corroborates authors’ earlier findings by TG/MS that dehydration is the most significant

reaction during pyrolysis157

. Also, it can be observed that the two small peaks at

approximately 1510 and 1460 cm-1

which are related to the Guaiacyl and Syringyl units in the

LS200

, are much less pronounced after pyrolysis at 400°C, indicating the thermal degradation

of the former LS material. These two peaks continue to vanish upon heating to higher

temperatures. Moreover, aromatic condensation becomes more and more visible for chars

carbonized beyond 600°C. This aromatic condensation is accompanied with a decline of the

aliphatic C-H groups (3000-2800cm-1

), the net decrease of the alkyl sulfonate functionalities

(around 1150 cm-1

) and C-O functionality (around 1030 cm-1

).Besides, the broad band with

some sub-bands between 1600-1000cm-1

(including the in-plane C-H bending vibrations that

interact with various aromatic ring C-C vibrations198,201

) decreases substantially with the HTT,

indicating the progressive aromatization of the formed carbonaceous material. Further

evidence of the aromatic ring condensation, as mentioned by several authors198,202

but not

quite obvious in the current results, is the appearance of three bands between 900-700cm-1

,

assigned to the out-of-plane C-H bending vibrations. For chars obtained from high

temperature pyrolysis, ca. 1000°C and 1200°C, all characteristic signals are largely reduced,

indicating the loss of mostly functional groups as well as the start of graphitization since

84

graphite has no characteristic infrared bands in the investigated wavenumber range (spectrum

of graphite could be found in articles201,203

) and the FTIR is not sensitive to long-range

ordering during graphitization204

.

Fig 3.3.2 FTIR spectra of MFC/LS composites before and after pyrolysis to different HTTs

Generally speaking, pyrolysis up to 400-1200°C generates a continuous reduction of

functional groups in chars, mainly due to the dehydration199

and aromatic

condensation150,199,201

. Over 800°C, only some oxygen containing functional groups resist

from the pyrolysis and most of the carbon content appears to be incorporated into condensed

aromatic structures, making its FTIR spectrum close to that of polycrystalline graphite

3.3.4 Density evolution

Fig 3.3.3 exhibits both bulk and skeletal (true) densities of MFC/LS sheet and those of its

derived carbons. Skeletal densities rise from 1.45 to 2.05 for HTT ranged from 400 to 800°C,

which is in good agreement with the literature205–208

. Such increase in true densities is thus

considered as reflecting the continuous synthesis of highly organized carbon structures

(turbostratic structures) that are more compact and denser than disordered carbon205,206

.

During pyrolysis, non-conjugated molecules either decompose as volatiles or convert into

conjugated ones and the chars undergo aromatic condensation. Due to the existence of defects

(oxygen components that are heat-resistant) and the randomly oriented graphite crystals

(turbostratic nature), the maximum skeletal density of biochars is always below that of

graphite (2.25 g/cm3)205,208

and within the range of 2-2.1 g/cm3.206,208

However, a slight drop

of skeletal density to 1.79 g/cm3 is observed for the sample carbonized at 1000°C which is

later followed by a recover up to 1.94 g/cm3 (1200°C). To authors’ best knowledge, such

60080010001200140016001800200022002400260028003000320034003600380040000.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

[cm-1

]

Nor

mal

ized

Abs

orba

nce

[a.u

]

before pyrolysis

400

600

800

1000

1200

bonded O-H stretch aliphaticC-H

aromaticC=C

in-plane C-H bendingsinteracting with

various ring vibrations

alkylsulfonates

C-Oout-of-planeC-H bending

85

irregular changes of skeletal density were rarely reported elsewhere and should be associated

with the departure of heteroatoms, probably oxygen, while the carbonaceous char is suffering

from structural rearrangement.

Fig 3.3.3 Bulk and skeletal densities of MFC/LS precursor (23°C) and carbons as well as the

calculated porosity

Bulk density of biocarbons is the result of the competition between the HTT-induced weight

loss and volumetric shrinkage. It generally follows the evolution profile of the skeletal density

although with a lower value because of the existence of the porosity. Bulk densities slightly

increase from 1.30 to 1.48 g/cm3 with elevated HTTs until 800°C (Fig 3.3.3), quite similar to

those of biochars as reported elsewhere156,205,207

. This increase of bulk density is the result of

the weight loss that is less important than the volumetric shrinkage throughout this

temperature range207

. Above 800°C, bulk density falls immediately into 1.14-1.18 g/cm3, as a

result of the continuous loss of weight whereas the dimensional shrinkage is quite small207

.

The calculated porosities also increases with HTT (Fig 3.3.3), which is in agreement with the

previous SEM observation (Fig 3.3.1) and the literature206,207

. Compared to wood and grass

precursor, chars derived from MFC/LS precursor have significantly lower porosities, ranged

from 0.05 to 0.4, corresponding to HTT at 400°C and 1200°C (over 0.6 for wood and grass

carbons). This difference was associated to the presence of a continuous LS matrix in the

pristine MFC/LS composite.

3.3.5 Electrical conductivity evolution

The evolution of biochars’ bulk conductivity as a function of HTT is presented in Fig 3.3.4.

According to the measurement, chars of MFC/LS sheets begin to achieve a DC measureable

electrical conductivity after carbonization at around 600°C. An extremely important increase

of more than 3 orders of magnitude occurs between 600 and 700°C, from 4.86 × 10−4 to 1.29

86

S/cm, as reported by many other studies concerning biomass pyrolysis10,156,168,201

. Between

700 and 1000°C, the DC conductivities of MFC/LS chars rise linearly with the heating

temperatures, indicating the strongly HTT-dependent conductivity of as-produced carbons.

After 1000°C, the temperature effect on electrical conductivity levels off.

Fig 3.3.4 Evolution of electrical conductivity of chars as a function of HTT

3.3.6 Microstructural evolution

3.3.6.1 Raman analysis

Since Raman spectroscopy is particularly sensitive to sp2 carbon structures and their features

measured on the scale of nanometers167,209

, it is widely used to evaluate the development of

graphite-like microstructures in disordered or turbostratic carbon materials210–213

. Normalized

Raman spectra of as-carbonized chars are demonstrated in Fig 3.3.5 and the intensities of the

most characteristic bands are summarized in Fig 3.3.6. According to Fig 3.3.6, the D band

(located approximately at 1300 cm-1

) is visibly pronounced, from 0.2064 to 0.4163 when the

HTT was increased from 400 to 1200°C. The G band (1600 cm-1

) is only slightly enhanced

with increasing HTTs. On the contrary, the V band (at around 1500 cm-1

) together with S

band (1200 cm-1

) and Gr band (1800 cm-1

) are continuously reduced from over 0.7 to below

0.5. Similar evolution of Raman bands was given in the literature163,165,214,215

. The most

important modifications of bands occur between 400 and 600°C for which the band V, S and

Gr are importantly reduced whereas the D and G band are visibly enhanced (more visibly in

Fig 3.3.5). Such band changes are supposed to be correlated with the transformation from

insulators to conductors for the chars carbonized at 400 and 600°C.

87

Fig 3.3.5 Raman spectra of MFC/LS chars prepared at 400 to 1000°C, normalized to G band

positions

Fig 3.3.6 Calculated area intensities of band D, G, (S+V+Gr) as well as D/G band ratio as

function of HTTs

800 1000 1200 1400 1600 1800 20000

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Raman shift (cm-1

)

Inte

nsity (

A.U

.)

400

600

700

800

900

1000

1200

400 500 600 700 800 900 1000 1100 12000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

HTT °C

Specific

band a

rea inte

nsity (

A.U

)

400 500 600 700 800 900 1000 1100 12002.4

2.6

2.8

3

3.2

3.4

3.6

3.8

4

I D/I

G r

atio (

A.U

)

ID

/Itotal

IG

/Itotal

(IV+S+Gr

)/Itotal

ID

/IG

88

As pointed out in the literature10,150,211

, G band arises from the sp2 bonded carbon structures

and is intensively present in highly ordered carbon structures like graphite whereas D band is

attributed to the disorder around sp2 carbon, such as defects, distortions and the edges of

graphite crystals. Therefore, the band ratio of D to G is often used to characterize the structure

ordering in carbon materials168

. An increase of the band area of D over G could be obtained

according to the deconvolution results, as shown in Fig 3.3.6. Such an increase reveals a more

important rise of defect or edge-bonded graphite-like structures compared to that of

graphitization, indicating an increase in the concentrations and/or the size of aromatic

clusters164,166

. More specifically, due to the oxygen-rich precursor as was used in this work,

defects, mainly composed of oxygen, are inevitable since oxygen remains even after high

temperature pyrolysis (according to FTIR results in Fig 3.3.2). Although D band was named

after defects, its appearance does require the presence of ordered carbon structures (≥6

benzene rings) in close proximity164,166,167

.Furthermore, V, S and Gr bands, which are related

to the presence of amorphous carbon structures167

decrease at elevated HTTs. This would

imply a continuous removal of amorphous carbon structures during pyrolysis through

devolatilization and/or conversion of the amorphous structures into more ordered graphite-

like ones167,216

.

It is worth noticing that the D/G ratio is observed to decrease in the bio-carbon obtained after

pyrolysis at 1200°C (Fig 3.3.6). The existence of a maximum of the D/G ratio in the

temperature range from 580°C to 1500°C is commonly observed in the related

studies10,164,212,217

. With the enlargement of the aromatic ring system that is further enhanced

by temperatures over 1000°C, the decrease of the “defects” (band D) with regards to an

important growth of regular graphene stacks (band G) is supposed to lead to a decrease of

D/G band ratio164

. The reduction of defects (oxygen-containing) could also be confirmed from

FTIR results in Fig 3.3.2 for the sample of 1200°C.

89

Fig 3.3.7 Band position variations of chars carbonized at different HTTs

In addition to band ratios, band positions could also provide some clues about chars’

microstructural modifications. In Fig 3.3.7, G band is found to shift to higher frequencies

(blue shift) with elevated HTTs and in contrast to that, D band tends to move towards lower

positions (red shift) while V band do not show a noticeable position change. It is believed that

G band position is strain-related218–220

. As aromatic clusters grow larger and more graphite-

like during heat treatment, they begin to collide, impinge and/or merge with neighboring

clusters within a confined volume by putting themselves under increasing compression167

.

Therefore, such blue shift might be another sign for the growth of ordered carbon clusters. On

one hand, D band position red shifting with increasing HTTs could be assumed to be due to

the ever increasing tensile strain that the near-edge structures suffer from, in contrast to the

blue shift of G band220

. On the other hand, D band position allows providing information

about the most common types of near edge structures around the ordered carbons167

. For chars

prepared at 400°C, their D band is located at around 1350 cm-1

, corresponding to that of laser-

reduced graphene oxide whose predominant edge structures are amorphous carbon bonded

(sp3 bonded). It also collaborates with the fact that 400°C carbonized char has the most

important amorphous carbon content. As HTT increases to 900°C, D band of chars moves

gradually towards the D band position of synthetic graphite (around 1316 cm-1

)167

, indicating

that the edges become similar to those in graphite. Then at 1000°C, the curvature of the

edges/defects is pronounced and becomes more fullerene-like, which leads its D band position

to approach 1300 cm-1

.221

3.3.6.2 XRD analysis

XRD has been widely used to help estimating the dimensions of graphite –like crystallites in

turbostratic carbon materials8,222–225

. The diffractograms of as-carbonized chars are displayed

200 400 600 800 1000 1200 14001250

1300

1350

1400

1450

1500

1550

1600

HTT °C

Band p

ositio

ns c

m-1

G band

V band

D band

90

in Fig 3.3.8. In the current measuring range, all diffractograms exhibit a broad (002) reflection

at 2𝜃 ≈25° and a dimensional lattice band (100) at 44°, indicating the existence of graphene-

layer stacks in the chars150,223,225

. The interlayer spacing d002, stack thickness Lc and the lateral

size of stacks La were as-calculated and shown in Table 3.3.1.

Fig 3.3.8 Normalized XRD diffractograms of chars a) before and b) after linear-subtraction

Table 3.3.1 Crystal parameters in MFC/LS carbons

HTT (°C) 400 600 700 800 900 1000

d002 (nm) 0.38 0.39 0.38 0.37 0.37 0.38

Lc (nm) 0.74 0.86 0.87 0.75 0.75 0.86

La (nm) 2.14 2.49 2.78 3.14 3.34 3.40

N 2.95 3.23 3.29 3.02 3.02 3.24

A general rise of lateral size of crystallites, from 2.135 nm to 3.395 nm can be observed when

the carbonization temperature increases from 400 to 1000°C. Quite similar values were

reported in pitch-based carbon fibers212

and pine wood chars159

. The interlayer spacing values

of MFC/LS chars fall in the typical range of 0.37-0.41 nm for cellulose-based chars

carbonized between 400 and 1000°C158,199,201

. The thickness of the stacks of graphene layers

varies between 0.74-0.89 nm, signifying that each stack consists mainly of 3 graphene sheets

(the number of layers was calculated according to Eq 2.6.4). The resulting crystal parameters

are displayed in Table 3.3.1, which are in good agreement with the data provided in the

literature150,201

.

0 10 20 30 40 50 60 700.5

1

1.5

2

2.5

3

3.5

4

2 (°)

Inte

nsity (

A.U

.)

400

600

700

800

900

1000

0 10 20 30 40 50 60-0.5

0

0.5

1

1.5

2

2.5

2 (°)

Inte

nsity (

A.U

.)

400

600

700

800

900

1000

a) b)

91

3.3.7 Discussion

3.3.7.1 Development of electrical conductivity during pyrolysis: impact of porosity

As highlighted by the SEM images (Fig 3.3.1), MFC/LS carbons present stable bulk

structures consisting of a hard carbon matrix and crack-like pores. The conductivity of the

bulk structure is thus supposed to be reduced by the presence of porosity. In order to eliminate

the porosity effect, skeletal conductivity that represents only the conductivity of the hard

carbon was calculated according to Eq 3.3.1. Such expression is based on an open carbon

foam model226

, where 𝜎 and 𝜌 represent the conductivity and the density of the material,

respectively, and the subscript s denotes skeletal properties (hard carbon) whereas the terms

without it indicate bulk foam properties.

Eq 3.3.1 𝜎

𝜎𝑠=

1

3(

𝜌

𝜌𝑠) +

2

3(

𝜌

𝜌𝑠)1.5

As expected in Fig 3.3.9, the skeletal conductivities display higher values than bulk ones but

roughly follow the profile of the latter with regards to HTTs. After carbonization at 1200°C,

MFC/LS biochar could achieve an electrical conductivity as high as 186 S/cm, which is about

one order of magnitude lower than that of polycrystalline graphite (1250 S/cm)18,156

and in

line with the presence of defect or edge-bonded graphite-like structures after carbonization at

1200°C.

Fig 3.3.9 Skeletal conductivity as function of HTTs

3.3.7.2 Development of electrical conductivity during pyrolysis: impact of microstructures

and HTTs

Leaving aside the porosity effect, the conductivity of MFC/LS hard carbon is still strongly

HTT-dependent (Fig 3.3.9) and must be correlated to the mutual evolution of the conductive

and insulating phases in the chars10,150

. McDonald-Wharry et al.150

proposed a

phenomenological model for illustrating the microstructural changes during pyrolysis for non-

graphitizing precursors. Based on this model as well as the present results, a model describing

92

the electrical conductivity development in MFC/LS carbons was derived, as show in Fig

3.3.10.

By combining the results concerning microstructure evolution with that of electrical property,

the conductivity development of MFC/LS carbons is supposed to pass through several

different regimes as function of HTTs:

i. 400°C ≤ HTTs≤ 600°C: chars at this period already possess regular graphene stacks in

small size (2,135 nm), as evidenced by XRD results. However, they are completely

isolated by defected graphene structures (detected by Raman band D) and amorphous

structures (Raman band V, S and Gr) with the latter far more importantly present in

the chars. Thereby, electron hopping and tunneling is prevented or largely reduced.

Consequantly, DC conductivity of chars is undetectable or quite low (Fig 3.3.10a)).

ii. 600°C <HTTs< 700°C: with the amorphous phase continuously reducing or

incorporating into structured phases, the latter, regardless of regular graphene stacks or

defected ones, is growing and approaching to each other. Upon a percolation threshold,

they get in mutual contact thus largely favoring electrons’ movement. In hence, the

DC conductivity is sharply improved. Since the defected graphene phase is far more

important than the regular one in MFC/LS carbons, which is evidenced by Raman D/G

ratio that is always more than 1, it is supposed that the electrical pathways are

constructed more probably by contacts between defected graphene clusters (Fig

3.3.10b)).

iii. 700°C ≤HTTs≤ 1000°C: during this period, both regular and defected graphene

structures continue to grow in size and in number, intensifying the mutual contacts

between conductive clusters. As a result, electrical conductivity is strengthened (Fig

3.3.10c)).

iv. HTTs>1000°C: once the current pathway has been established, further graphitization

or reduction of defects will not lead to important increase in contacts between

conductive phases, thus the HTT-induced development of electrical conductivity

slowed down (Fig 3.3.10d)).

Fig 3.3.11 displays the relationship between the skeletal conductivity of MFC/LS carbons and

the Raman ratio of band D and G over the total band area as well as graphene crystal size La.

It could be found that the conductivity is quite positively correlated to the crystal size and the

D+G band ratio, confirming that the conductivity improves with strong relationship to the

growth of both regular graphene stacks’ and defected graphene structures. Furthermore, since

MFC/LS carbons have defected-graphene-dominated structure (ID/IG is always more than 1),

the growth of the defected graphene phase is more significant, compared to regular one, in

terms of ensuring the conductivity of the resulting hard carbon.

93

Fig 3.3.10 Illustrative schema for describing the conductivity development in MFC/LS

carbons

Fig 3.3.11 Correlation between skeletal conductivities and band ratio ID+G/Itotal as well as

crystal size La

3.3.7.3 Comparison of biocarbons from various precursors in terms of electrical property

After one step casting and carbonization at 1000°C, MFC/LS sheets could achieve a

surprisingly high electrical conductivity of 95 S/cm with a relatively low density 1.14 g/cm3,

even compared to polycrystalline graphite (1250 S/cm and 1.6 g/cm3)18

. A comparison with

other biocarbons in terms of physical properties was made and presented in Table 3.3.2.

94

Although different compositions and physical forms that make it hard to compare directly the

specific electrical conductivity between the listed biocarbons, authors can still remark the

mutual promotion between MFC and LS in terms of carbon quality: i) compared to MCC

carbon, MFC/LS contains also micro-scaled cellulose fibers but could achieve nearly double

conductivity at the same carbonization temperature mainly because of its LS content that is

heat-resistant and contributes significantly the carbon yield so as the electrical property; ii)

without MFC that constructed a micro-scaled web that tightly cross-linked LS molecules, LS

solely could not form self-standing sheet that is later thermally stable.

Table 3.3.2 Comparison of biocarbons from various precursors

Ref

Bulk

conductivity

(S/cm)

Bulk density

(g/cm3)

Physical form Precursor

This work 95 (1000°C) 1.14 Self-standing

sheet MFC/LS blend

Rhim et al.10

50 (1000°C) 1.17 Bulk pellet Microcrystalline

cellulose (MCC)

Kercher et

al.156

20 (1000°C) 0.595 Monolith Fiberboard

Deraman et

al.180

4.9 (1000°C) 1.05 Bulk pellet

Lignin/precarbonized

carbon

Lu et al.181a

0.9-20 0.065-0.6 Monolithic

aerogel

Polycondensed

resorcinol-

formaldehyde

Sanchez-

Gonzalez et

al.182a

0.44-2.15 0.25-0.7 Compressed

carbon black Carbon black

a Carbons from precursors other than biomass, given for comparison

3.3.8 Conclusions

This part of work characterized the biocarbons elaborated by slow pyrolysis of MFC/LS

precursors. A wide range of heat treatment temperatures (HTTs: 400-1200°C) was explored

for biocarbons’ synthesis in order to better understand the improvement of their

morphological, physicochemical and microstructural properties, and the most importantly, the

electrical properties. MFC/LS carbons have been proved to be heat-resistant and

morphologically stable to slow pyrolysis. FTIR results provided evidences concerning the loss

of the majority of functional groups during heat treatment in MFC/LS carbons and only a few

oxygen-containing structures were retained from high-temperature pyrolysis. Microstructural

95

characterization by Raman revealed that the resulting carbons are composed of i) defect-free

(regular) graphene stacks (3 layers of graphene, according to XRD); ii) defected (mainly

oxygen) graphene structures and iii) amorphous phases. XRD results allow to perceiving

directly the growth of regular graphene stacks with elevated HTTs whereas Raman ratios help

understanding the relative changes of all three phases. As a result of the growing of

conductive phases (regular and defected) in addition to the reduction of the insulating ones

(amorphous), a general increase of electrical conductivity was observed in MFC/LS hard

carbons as function of HTTs although the mechanisms for the conductivity development are

not the same in different HTT ranges. Based on that, a model describing the electrical

conductivity development in MFC/LS hard carbons was proposed. By accounting the porosity

effect, the bulk conductivities of MFC/LS carbons are also strongly HTT-enhanced which are

resulted from the competition between the conductivity strengthening in hard carbons and the

porosity development.

After carbonization at 1000°C, biocarbons from MFC/LS precursors are still self-standing

sheets and could achieve quite interesting electrical conductivity (95 S/cm) compared to their

relatively low density of 1.14 g/cm3. The resulting physical properties are really outstanding

in comparison with previously reported biocarbons.

96

3.4 Optimization of the formulation of carbon precursors for 3D printing

and carbon quality

3.4.1 Introduction

The work in section 3.1 reveals that add of LS into MFC hydrogels leads to either a disruption

of the hydrogels’ stability after 3D printing due to lack of viscosity (with 10%-30% LS), or a

loss of shape definition because of the long restauration time (with 50% LS). Nevertheless,

LS is an essential component to the composite since it largely contributes to the carbon yield

after pyrolysis which is important to ensure the conductivity of as-elaborated bio-carbons.

One solution to improving the hydrogels’ printability while keeping using large quantity of

LS in the formulation is to add appropriate additives. In the frame of this work that is to use

bio-sourced materials, cellulose powder was chosen.

In section 3.4, the impact of cellulose powder added into the formulation of MFC/LS blends

was investigated in terms of rheology and the resulting 3D printing, using the same

methodology as described in section 3.1. Furthermore, the pyrolysis procedure of

MFC/LS/CP composite was monitored using TGA, referring to section 3.2. Carbons

elaborated from MFC/LS/CP precursor were also compared to those from MFC/LS precursors

according to the microstructure, density, conductivity (section 3.3). In addition to that,

MFC/LS/CP carbons were characterized especially in mechanical strength.

3.4.2 Rheological properties of MFC/LS/CP hydrogel

The improvement of rheological properties by the addition of CP is shown in Fig 3.4.1. The

viscosity still keeps the shear-thinning profile but is strengthened by 2-3 magnitudes due to 22%

of CP (w/w+water) added into the pristine MFC/LS hydrogel. More importantly, the time for

network restauration is largely shortened, from 27s to 4s (since MFC/LS/CP is quite similar to

pure MFC behavior)15, with the presence of CP. The printing results (Fig 3.4.1a)) confirmed

the usability of as-formulated hydrogel.

Fig 3.4.1 Viscosity a) and b) thixotropic property of the MFC/LS/CP hydrogel, compared to

the corresponding binary one (MFC/LS)

a) b)

97

3.4.3 Macro- and micro-morphology of MFC/LS/CP precursor and the resulting

carbons

The photographs of 3D-printed MFC/LS/CP precursor composites are displayed in Fig 3.4.2a).

After air-drying at ambient temperature, the dry objects still maintain their original shape and

definition without visible defects due to the evaporation of water. Carbons derived from their

corresponding precursor are surprisingly well-shaped and also without major surface and

sectional irregularities although a rather important shrinkage occurred during the pyrolysis

(Fig 3.4.2b)). As a result, the excellent morphological stability of the MFC/LS/CP carbon

precursor could be concluded, that is the cornerstone of further characterizations of as-

elaborated carbons.

Fig 3.4.2 Photographs of: a) 3D-printed objects from MFC/LS/CP hydrogel after air-drying; b)

after pyrolysis at 1000°C

The micro-scaled morphology of the samples is presented in Fig 3.4.3 using a SEM. The used

cellulose powder (CP) particles (Fig 3.4.3a)) have an average length of 50µm and look like

pieces cut from the much longer cotton linters. Its presence in the MFC/LS/CP composite can

be clearly seen from the surface of the latter (Fig 3.4.3.b)). Such composite has a typically

“almond chocolate” construction in which CP is more like the tiny pieces of almond wrapped

by the “chocolate” surrounding that is made of MFC/LS. Since MFC was included in the

composite in very small quantity and also because of its tiny dimension, it is less visualized

than LS in the “chocolate-like” MFC/LS background. Furthermore, pyrolysis leads to limited

changes on the sample surface except largely broadening and intensifying the fissures among

which some seem already existed after drying (Fig 3.4.3b) & c)). Besides, it “smooths” the

surface due to the low residue of CP at 1000°C (confirmed by Fig 3.4.4c)). However, images

of samples’ section (Fig 3.4.3d) & e)) reveal an important appearance of macro-pores that is

very likely to be the result of CP decomposition during pyrolysis by accounting the

dimensions of pores (multi-diameters, up to 50µm) as well as the fact that CP was wrapped

randomly in three-dimension by the MFC/LS surrounding. The significant improvement of

a)

b)

98

the porosity by the addition of CP in the formulation (compared to MFC/LS composite) will

lead to a more promising use of the studied bio-carbons in energy storage devices. In addition

to that, it is worth noticing that although the sample was constructed with printed

filaments(diameter=500µm), no single filament was identified from the section image in

micro-scale (Fig 3.4.3d)), signifying that all filament were molten together after deposition

and the samples is rather homogeneous in terms of the distribution of the matter.

Fig 3.4.3 SEM observation of: a) as-received cellulose powder (CP); b) &c) surface of

MFC/LS/CP precursor composite and its carbon after pyrolysis at 1000°C, respectively; d)&e)

transversal section of samples corresponding to b) and c), respectively

3.4.4 Analysis of pyrolytic process of MFC/LS/CP composite

The TGA and DTG patterns of the tri-component composite are presented in Fig 3.4.4a) & b)

and are compared to those of its constituents: cellulose powder and the binary composite of

MFC/LS. CP exhibits a typical cellulose degradation peak at around 325°C with the

maximum reaction rate reaching 13.2%/min (or 2.64%/°C). Its thermal degradation profile

shows no difference compared to the published one in 123 where CP of the same origin

a) b) c)

d) e)

CP Before pyrolysis 1000°C

Before pyrolysis 1000°C

99

(Aldrich, 50µm) was examined. One remarkable phenomenon is that the used cellulose

powder (powder form) decomposes much more intensively than microfibrillated cellulose (in

the form of a fibrillated web), 13.2%/min compared to 6.1%/min and reaches the maximum

reaction rate at a slightly lower temperature (325°C with regard to 334°C, respectively), as

shown in Fig 3.4.4c). MFC is composed of smaller cellulose fibrils ( L≈10 µm, D≈2-50 nm12)

than CP. Due to the huge quantity of inter-fibril hydrogen bonds that construct the MFC

“web”, MFC decomposition is more or less inhibited227–229. As a result, MFC is thermally

more stable than the powder formed CP. However, in the range of 230°C-300°C, MFC

degrades slightly stronger than CP, probably due to its hemicellulose component230 whose

decomposition normally occurs at this temperatures124. In addition to that, many other factors

may also affect thermal degradation of cellulosic products, such as cellulose source231,

bleaching degree232, crystallinity233 and impurities234.

Fig 3.4.4 TG (a)) and DTG (b)) curves of the composite MFC/LS/CP and its constituents

(cellulose powder and the composite MFC/LS); c) a comparison of thermal degradation

between CP and MFC

Furthermore, it is remarkable that the triplet is rather thermally stable by generally following

the thermal profile of its most important constituent, the binary composite, except for the

region between 200°C and 280°C. Throughout this temperature range, the reaction rate of the

a) b)

c)

100

triplet is slightly higher than the binary composite, possibly due to the proportion of CP

whose decomposition was catalyzed by the presence of Na from the binary composite by

shifting to lower temperatures, as was proved in an earlier study157. The temperature shift of

CP could be quantified from Fig 3.4.5a) where the theoretically calculated curve assuming an

additivity law is obtained using Eq 3.4.1. The sodium-induced temperature shift of cellulose

powder is consistent with the previously determined linear relationship describing how

sodium content influences MFC decomposition in MFC/LS composites157

, as shown in Fig

3.4.5b).

Fig 3.4.5 a) experimental and calculated DTG curves of the ternary composite MFC/LS/CP; b)

the relationship of ΔTp with regard to the sodium content (the point representing the triple

composite is indicated by a flash whereas the others represent MFC/LS composites

Eq 3.4.1 𝐷𝑇𝐺(𝑇) 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 = % 𝐶𝑃 × 𝐷𝑇𝐺(𝑇)𝐶𝑃 + %𝑏𝑖𝑛𝑎𝑟𝑦 𝑐𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑒 × 𝐷𝑇𝐺(𝑇)𝑏𝑖𝑛𝑎𝑟𝑦 𝑐𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑒

3.4.5 Characterization of MFC/LS/CP carbons

3.4.5.1 Density evolution

Fig 3.4.6 exhibits both bulk and skeletal (true) densities of printed MFC/LS/CP “disk” and

those of its derived carbons. According to it, the skeletal density of MFC/LS/CP carbon

precursor is only slightly higher than the bulk one, 1.53 compared to 1.31 g/cm3. It should be

attributed to the low porosity that exists in such composite, ca. 0.146, induced by air-drying

(as shown in Fig 3.4.3). During pyrolysis when HTT increases from 400 to 1200°C, quite

different evolutions are observed for skeletal and bulk density.

a) b)

101

Fig 3.4.6 Density and porosity evolution of MFC/LS/CP carbons

Skeletal density of the obtained carbons globally rises from 1.43 to 1.84 g/cm3 for a HTT

increase from 400 to 1200°C, except for 900 and 1000°C, for which a slight density drop is

observed. As commonly reported in the literature205–208

, such an increase of skeletal density

with HTTs is considered as reflecting the continuous synthesis of highly organized carbon

structures (turbostratic structures) that are more compact and denser than disordered

carbon205,206

. During pyrolysis, non-conjugated molecules either decompose as volatiles or

convert into conjugated ones and the chars undergo aromatic condensation. Due to the

existence of defects (oxygen components that are heat-resistant) and the randomly oriented

graphite crystals (turbostratic nature), the maximum skeletal density of biochars is always

below that of graphite (2.25 g/cm3)205,208

.

Bulk density of biocarbons is the result of the competition between the HTT-induced weight

loss and volumetric shrinkage. Bulk density importantly drops from 1.31 to 0.77 g/cm3 when

HTT increased at 400°C. After 400°C, the bulk densities of MFC/LS/CP carbons fluctuate

within the range of 0.68 and 0.73 g/cm3, which seems to be less HTT-influenced.

The calculated porosities also increases with HTTs (Fig 3.4.6) at early stage of pyrolysis

before stabilizing after 600°C, which is in agreement with the previous SEM observation (Fig

3.4.3) and the literature206,207

. Compared to MFC/LS carbons, those derived from MFC/LS/CP

precursor have significantly higher porosities, ca. 0.6 with regard to 0.37 at 1200°C. This

value is approaching the result that was reported for wood chars205

. The distinguished porosity

between MFC/LS and MFC/LS/CP carbons is associated with the presence of CP in the

formulation of the latter. According to its thermal degradation pattern in Fig 3.4.4c) as well as

the SEM observation in Fig 3.4.3e), CP nearly disappears after 600°C (carbon yield <5%)

thus leading to the formation of the pores whose dimensions are visibly similar to those of CP

102

particles. Thereby, a controlled porosity in the resulting carbons could be expected by simply

a good dosage of CP quantity in the formulation of carbon precursor.

3.4.5.2 Microstructural evolution

The mutual evolution of the 3 identified phases (defect-free and defected graphene stacks as

well as amorphous structures)150

in the as-elaborated carbons provides essential information to

understand the changes of their electrical and mechanical properties. As shown in Fig 3.4.7

and Table 3.4.1, both carbons present similar profiles in Raman spectra and XRD

diffractograms, owing to their same fundamental constituents. Nevertheless, the small

difference in proportions between LS-based precursor and cellulose-based precursors (MFC

and CP) as well as their different aromatic condensation mechanisms during pyrolysis still led

to a slight distinction in microstructural parameters as shown in Table 3.4.1. At 1000°C,

MFC/LS/CP carbon contains less LS, 76.6% compared to 84% in MFC/LS carbon, which

resulted the smaller size of graphene stacks and higher content of ordered structures (D+G

ratio) although to a very limited degree.

Fig 3.4.7 a) Raman spectrum and b) XRD diffractogram of 1000°C-synthesed MFC/LS/CP

carbon, compared to the corresponding MFC/LS carbon

Table 3.4.1 Comparison of Raman and XRD characteristics between MFC/LS/CP and

MFC/LS carbon of 1000°C

Precursor

LS

proportion

(%)

d002

(nm) Lc (nm) La (nm) N

Raman band

ID+G/Itotal

MFC:LS:CP=1:49:14 76.6 0.37 0.81 3.22 3.2 0.4768

MFC:LS=1:5.25 84 0.38 0.86 3.39 3.2 0.5047

3.4.5.3 Electrical conductivity

The bulk and the calculated skeletal conductivities (Eq 3.3.1) of MFC/LS/CP carbons are

displayed in Fig 3.4.8. According to it, both skeletal and bulk conductivity of MFC/LS/CP

carbons are highly HTT-depended and present 3 different regions as deeply interpreted in

section 3.3.6.2: i) the appearance of mutual contacts between conductive carbon clusters when

800 1000 1200 1400 1600 1800 20000

0.5

1

1.5

2

Raman shift (cm-1

)

Inte

nsity (

A.U

.)

MFC/LS/CP 1000°C

MFC/LS 1000°C

10 20 30 40 50-0.5

0

0.5

1

1.5

2

2.5

2thelta (°)

Inte

nsity (

A.U

.)

MFC/LS/CP 1000°C

MFC/LS 1000°Ca) b)

103

600°C≤HTT≤700°C for which the skeletal electrical conductivity underdoes a sharp rise of 3

orders of magnitude, ca. from 0.001 to 1.054 S/cm; ii) the steady intensification of such

contacts which arises from the growth of the conductive clusters for 700≤HTT≤1000, leading

to a continuous improvement of the electrical conductivity from 1.054 to 164.58 S/cm ; iii)

the slowed-down increase of the conductivity due to the already established current pathways

that will be less-influence by the growth of conductive clusters for HTT>1000°C. Moreover,

the difference between the bulk and skeletal density that increases at elevated HTT confirms

the fact that the resulting carbons possess a more and more important porosity during

pyrolysis, as revealed by density measurement (Fig 3.4.6). At 1000°C, MFC/LS carbon has a

slightly higher skeletal conductivity than MFC/LS/CP carbon, 173.51 S/cm with regard to

164.58 S/cm. Such a difference collaborates with the previous observations in Table 3.4.1 that

MFC/LS owns slightly larger size of graphene stacks and higher structural ordering

(represented by D+G band ratio), that have been proved to be positively correlated with the

electrical conductivity that is developed in engineering carbons during pyrolysis (section

3.3.6.2).

Fig 3.4.8 Bulk and calculated skeletal electrical conductivity of MFC/LS/CP carbons at

corresponding HTTs (MFC/LS carbon at 1000°C given for comparison)

3.4.5.4 Mechanical properties

Pyrolysis-induced modifications of the mechanical strength of MFC/LS/CP samples were

characterized by 3-point bending test. Despite the occasionally important deviation which is

caused by the commonly reported heterogeneity2,160,199 of engineering carbons (due to

disordered or turbostratic carbon structures), the results shown in Fig 3.4.9 display a strongly

HTT-dependent mechanical behavior for as-produced carbons whether carbonized or not,

MFC/LS/CP are elasticity-dominated materials with no plastic zone except for non-

104

carbonized ones for which a short plastic zone just before rupture was observed. The slope,

known as elastic modulus Ef, and the peak stress from each stress-strain curve are exhibited as

function of HTTs in Fig 3.4.9a) and b).

MFC/LS/CP carbon precursor exhibits a flexural elastic modulus of 4.05 GPa and a peak

stress of 23.53 Mpa, according to Fig 3.4.9. Compared to the precursor, the carbonaceous char

obtained at 400°C lose much mechanical strength reflecting in both elastic modulus and peak

stress. It is associated with the intensive mass loss due to devolatilization that the char suffers

from during this temperature range. The diminished mechanical property also coincides with

the decrease of the skeletal and bulk density of MFC/LS/CP at 400°C (Fig 3.4.6). When HTT

increases to 600-900°C, the temperature-strengthened mechanical properties become visible.

Both flexural modulus and peak stress rise quite linearly with HTTs and reach their maximum

(6.62 GPA and 29.79 MPa, respectively) at 900°C. After 900°C, further carbonization is

found to lower the mechanical strength of the carbons.

Fig 3.4.9 a) Elastic modulus and b) peak stress of MFC/LS/CP monolines pyrolyzed at

different HTTs from 3-point bending tests; c) the flexural stress-strain profile of the samples;

d) calculated skeletal elastic modulus with comparison to bulk ones

105

A linear-elastic beam deflection model was used to approximate the flexural modulus of the

hard carbon of MFC/LS/CP156,235

, as shown in Eq 3.4.2:

Eq 3.4.2 𝐸

𝐸𝑠= (

𝜌

𝜌𝑠)2

Where 𝐸 and 𝐸𝑠 correspond to the bulk and skeletal elastic modulus of the carbon; 𝜌 and 𝜌𝑠

are respectively the bulk and skeletal density. The approximate Young’s modulus data of the

MFC/LS/CP hard carbon are displayed in Fig 3.4.9d). Because of the porosity effect, the

skeletal modulus is found to be higher than the bulk one and the difference between them is

largely pronounced throughout the HTT range of 600-1200°C where the porosity is also

intensively present. The resulting hard carbon achieves the best Young’s modulus as high as

38.4 GPa at 900°C, that is nearly 1.5 higher than that of polycrystalline graphite

(27GPa156,235

).

3.4.6 Conclusions

The last part of the thesis examined the feasibility of manufacturing electrically conductive

and mechanically resistant carbon structures by 3D printing and subsequent pyrolysis using

MFC/LS/CP formulation. MFC, LS and CP are purely bio-sourced materials. Their water-

processing and interesting rheological properties (high viscosity, shear-thinning profile and

the short relaxation duration) account for the success of manufacturing well-shaped hydrogels

by 3D printing. The printed MFC/LS/CP hydrogels were proved to be morphologically stable

to air drying and to the subsequent pyrolysis. Owing to the presence of CP that is thermally

less-stable, the resulting carbons exhibit improved porosity at elevated HTTs, compared to

MFC/LS carbons. Such an increase in porosity is supposed to promote the use of MFC/LS/CP

carbons in energy storage devices. At 900°C, MFC/LS/CP carbons not only result in a high

electrical conductivity of 47.8 S/cm together with a low bulk density of 0.74 g/cm3 as well as

an important porosity of 0.58, but also achieve an elastic modulus maximum of 6.62 Gpa. It is

believed by authors that the interesting electrical and mechanical results obtained will lead to

a promising application as electrode materials for MFC/LS/CP biocarbons in close future.

106

4. Conclusions and

perspectives

107

4 Conclusions and perspectives Nowadays, the ever increasing environmental issues require the valorization of green

materials in all domains. The current work elaborated biocarbons by slow pyrolysis from

MFC/LS and MFC/LS/CP precursors that are purely lignocellulosic materials. The studies

conducted in previous sections evaluate the usability of such blends as engineering carbon

precursors in terms of rheology, thermal and kinetic characterization as well as carbon

properties.

The study of rheological properties of MFC/LS hydrogels revealed that adding LS into MFC

hydrogels led to either a disruption of the hydrogels’ stability after 3D printing due to lack of

viscosity (with 10%-30% LS), or a loss of shape definition because of the long restauration

time (with 50% LS). The improved formulation by using CP as additive obtained not only a

strengthened viscosity by 2-3 magnitudes but also a largely shorted restoration duration from

27s to 4s while keeping the shear-thinning profile. The objects printed using MFC/LS/CP

formulation had good definition, which also collaborate with the rheological conclusions.

The TG/DTG analysis showed hierarchically thermal degradation patterns between LS, MFC

and CP. Cellulose degrades intensively within a narrow temperature range between 250°C

and 350°C, while LS decomposes much more gently within 2 main temperature ranges,

150°C-500°C and 600°C -800°C. However, despite the same chemical composition, CP

decomposes much more intensively than MFC due to lack of huge quantity of hydrogen

bonds. Furthermore, the catalytic effect of sodium has been confirmed in both MFC/LS and

MFC/LS/CP blends by a highly correlated relationship between Tp shift and sodium content.

MFC/LS biocarbons elaborated from a wide range of heat treatment temperatures (HTTs:

400-1200°C) were characterized in the morphological, physicochemical and microstructural

properties. They have been proved to be heat-resistant and morphologically stable to slow

pyrolysis. FTIR results provided evidences concerning the loss of the majority of functional

groups during heat treatment and only a few oxygen-containing structures were retained from

high-temperature pyrolysis. Microstructural characterization by Raman revealed that the

resulting carbons are composed of i) defect-free (regular) graphene stacks (3 layers of

graphene, according to XRD); ii) defected (mainly oxygen) graphene structures and iii)

amorphous phases. As a result of the growing of conductive phases (regular and defected) in

addition to the reduction of the insulating ones (amorphous), a general increase of electrical

conductivity was observed in MFC/LS hard carbons as function of HTTs although the

mechanisms for the conductivity development are not the same in different HTT ranges.

Based on that, a model describing the electrical conductivity development in MFC/LS hard

carbons was proposed. After carbonization at 1000°C, biocarbons from MFC/LS precursors

are still self-standing sheets and could achieve quite interesting electrical conductivity (95

S/cm) compared to their relatively low density 1.14 g/cm3. The resulting physical properties

are really outstanding in comparison with previously reported biocarbons.

108

Based on the previous conclusions, an improved formulation of using MFC/LS/CP blends

accounts for a final success of elaborating electrically conductive and mechanically resistant

carbon structures. Owing to the presence of CP that yields nearly nothing after 600°C, a

controllable porosity in the resulting carbons could be expected by simply a good dosage of

CP quantity in the formulation of carbon precursor. Such a controllable porosity is supposed

to promote the use of MFC/LS/CP carbons in energy storage devices. At 900°C, MFC/LS/CP

carbons not only result in a high electrical conductivity of 47.8 S/cm together with a low bulk

density of 0.74 g/cm3 as well as an important porosity of 0.58, but also achieve an elastic

modulus maximum of 6.62 Gpa.

Clearly, physical properties of engineering carbons could be optimized by the control of the

carbonization process and the carbon precursor formulation. Due to specific thermal

degradation patterns of each component, a controllable porosity could be resulted at different

temperature ranges. Moreover, a low heating rate is essential to obtain carbonaceous char with

a good morphological stability. It is believed by authors that the interesting electrical and

mechanical results obtained will lead to a promising application as electrode materials for

MFC/LS/CP biocarbons in close future. In all, it is believed that this work proposes a new

way to elaborate electrodes by 3D printing. It also verifies and could promote the use of

lignocellulosic materials as active materials in energy storage devices after simply a slow

pyrolysis process. The synthesized host material could promise the future electrodes a high

specific capacity and a good cycling safety owing to its controllable porosity and low density.

Furthermore, the experiences of using 3D printing in this work make it possible to produce

samples with flexible forms and dimensions in order to meet different characterization

purposes. In the scale of laboratory, small objects in the form of single line or thin film could

be elaborated for the corresponding mechanical or electrical characterization. From a point of

view of applications, larger objects could be manufactured after an appropriate formulation in

order to meet rheological requirements. More excitingly, electrodes of energy storage devices

(i.e. Li-ion battery) could be directly printed using a two-head printer with one head firstly

printing a host material then another filling with other types of hydrogel (probably

electrochemistry-strengthen components). As a result, 3D printing not only amplifies the

potential applications of MFC/LS/CP as a printable hydrogel but also proposes an unexplored

way to elaborate electrodes of next generation. In future, work will be focused on more

advanced formulations of 3D printing “ink” by including, for instance, UV-curing

components for fast manufacturing. Functional materials could also be formed by 3D printing

using carbon/metal formulations.

109

5. References

110

5 References 1. Pandolfo, A. G. & Hollenkamp, A. F. Carbon properties and their role in supercapacitors.

J. Power Sources 157, 11–27 (2006).

2. Burchell, T. D. Carbon Materials for Advanced Technologies. (Elsevier, 1999).

3. Lehmann, J. & Joseph, S. Biochar for Environmental Management: Science, Technology

and Implementation. (Routledge, 2015).

4. Liu, W.-J., Jiang, H. & Yu, H.-Q. Development of Biochar-Based Functional Materials:

Toward a Sustainable Platform Carbon Material. Chem. Rev. 115, 12251–12285 (2015).

5. Ippolito, J. A., Laird, D. A. & Busscher, W. J. Environmental Benefits of Biochar. J.

Environ. Qual. 41, 967–972 (2012).

6. Perepelkin, K. E. Carbon fibres with specific physical and physicochemical properties

based on hydrated cellulose and polyacrylonitrile precursors. A review. Fibre Chem. 34,

271–280 (2002).

7. Liu, W.-J., Tian, K., He, Y.-R., Jiang, H. & Yu, H.-Q. High-Yield Harvest of

Nanofibers/Mesoporous Carbon Composite by Pyrolysis of Waste Biomass and Its

Application for High Durability Electrochemical Energy Storage. Environ. Sci. Technol.

48, 13951–13959 (2014).

8. Jin, J., Yu, B., Shi, Z., Wang, C. & Chong, C. Lignin-based electrospun carbon

nanofibrous webs as free-standing and binder-free electrodes for sodium ion batteries. J.

Power Sources 272, 800–807 (2014).

9. Wang, Y., Zou, Y., Chen, J., Li, G.-D. & Xu, Y. A flexible and monolithic nanocomposite

aerogel of carbon nanofibers and crystalline titania: fabrication and applications. RSC Adv.

3, 24163 (2013).

10. Rhim, Y.-R. et al. Changes in electrical and microstructural properties of microcrystalline

cellulose as function of carbonization temperature. Carbon 48, 1012–1024 (2010).

11. Mun, S. P., Cai, Z. & Zhang, J. Fe-catalyzed thermal conversion of sodium lignosulfonate

to graphene. Mater. Lett. 100, 180–183 (2013).

12. Lavoine, N., Desloges, I., Dufresne, A. & Bras, J. Microfibrillated cellulose – Its barrier

properties and applications in cellulosic materials: A review. Carbohydr. Polym. 90, 735–

764 (2012).

13. González, I. et al. From paper to nanopaper: evolution of mechanical and physical

properties. Cellulose 21, 2599–2609 (2014).

14. Myrvold, B. O. A new model for the structure of lignosulphonates: Part 1. Behaviour in

dilute solutions. Ind. Crops Prod. 27, 214–219 (2008).

15. Shao, Y., Chaussy, D., Grosseau, P. & Beneventi, D. Use of Microfibrillated

Cellulose/Lignosulfonate Blends as Carbon Precursors: Impact of Hydrogel Rheology on

3D Printing. Ind. Eng. Chem. Res. (2015). doi:10.1021/acs.iecr.5b02763

16. Shao, Y., Guizani, C., Grosseau, P., Chaussy, D. & Beneventi, D. Thermal

characterization and kinetic analysis of microfibrillated cellulose/lignosulfonate blends. J.

Anal. Appl. Pyrolysis doi:10.1016/j.jaap.2017.03.001

17. Mauter, M. S. & Elimelech, M. Environmental Applications of Carbon-Based

Nanomaterials. Environ. Sci. Technol. 42, 5843–5859 (2008).

18. Michio, I. & Kang, F. Carbon Materials Science and Engineering: From Fundamentals

to Applications. (清华大学出版社有限公司, 2006).

19. Baker, D. A. & Rials, T. G. Recent advances in low-cost carbon fiber manufacture from

lignin. J. Appl. Polym. Sci. 130, 713–728 (2013).

20. Norberg, I. Carbon fibres from kraft lignin. (KTH Royal, 2012).

21. Dumitriu, S. Polysaccharides: Structural Diversity and Functional Versatility, Second

Edition. (CRC Press, 2004).

22. Higuchi, T. Biosynthesis and biodegradation of wood components. (Elsevier, 2012).

111

23. Barnett, J. R. & Bonham, V. A. Cellulose microfibril angle in the cell wall of wood fibres.

Biol. Rev. 79, 461–472 (2004).

24. Klemm, D., Heublein, B., Fink, H.-P. & Bohn, A. Cellulose: Fascinating Biopolymer and

Sustainable Raw Material. Angew. Chem. Int. Ed. 44, 3358–3393 (2005).

25. Habibi, Y., Lucia, L. A. & Rojas, O. J. Cellulose Nanocrystals: Chemistry, Self-Assembly,

and Applications. Chem. Rev. 110, 3479–3500 (2010).

26. Jarvis, M. Chemistry: Cellulose stacks up. Nature 426, 611–612 (2003).

27. Mazeau, K. & Heux, L. Molecular Dynamics Simulations of Bulk Native Crystalline and

Amorphous Structures of Cellulose. J. Phys. Chem. B 107, 2394–2403 (2003).

28. Azizi Samir, M. A. S., Alloin, F. & Dufresne, A. Review of Recent Research into

Cellulosic Whiskers, Their Properties and Their Application in Nanocomposite Field.

Biomacromolecules 6, 612–626 (2005).

29. Goring, D. A. I. & Timell, T. E. Molecular weight of native celluloses. Tappi 45, 454–460

(1962).

30. Pettersen, R. C. The chemical composition of wood. Chem. Solid Wood 207, 57–126

(1984).

31. Ebringerová, A. Structural Diversity and Application Potential of Hemicelluloses.

Macromol. Symp. 232, 1–12 (2005).

32. Belgacem, M. N. & Gandini, A. Monomers, Polymers and Composites from Renewable

Resources. (Elsevier, 2011).

33. Kadla, J. F. et al. Lignin-based carbon fibers for composite fiber applications. Carbon 40,

2913–2920 (2002).

34. Lora, J. H. & Glasser, W. G. Recent industrial applications of lignin: a sustainable

alternative to nonrenewable materials. J. Polym. Environ. 10, 39–48 (2002).

35. Hu, B. et al. Engineering Carbon Materials from the Hydrothermal Carbonization Process

of Biomass. Adv. Mater. 22, 813–828 (2010).

36. Wang, L., Schütz, C., Salazar-Alvarez, G. & Titirici, M.-M. Carbon aerogels from

bacterial nanocellulose as anodes for lithium ion batteries. RSC Adv. 4, 17549 (2014).

37. Tarascon, J.-M. & Armand, M. Issues and challenges facing rechargeable lithium

batteries. Nature 414, 359–367 (2001).

38. Dunn, B., Kamath, H. & Tarascon, J.-M. Electrical energy storage for the grid: a battery

of choices. Science 334, 928–935 (2011).

39. Schalkwijk, W. A. van. & Scrosati, B. Advances in lithium-ion batteries. (Kluwer

Academic/Plenum Publishers, 2002).

40. Jabbour, L., Bongiovanni, R., Chaussy, D., Gerbaldi, C. & Beneventi, D. Cellulose-based

Li-ion batteries: a review. Cellulose 20, 1523–1545 (2013).

41. Jabbour, L. Utilisation de procédés papetiers et de fibres cellulosiques pour l’élaboration

de batteries Li-ion Elaboration of Li-ion batteries using cellulose fibers and papermaking

techniques. (Université de Grenoble, 2012).

42. Wakihara, M. & Yamamoto, O. Lithium Ion Batteries: Fundamentals and Performance.

(John Wiley & Sons, 2008).

43. Araki, J., Wada, M., Kuga, S. & Okano, T. Flow properties of microcrystalline cellulose

suspension prepared by acid treatment of native cellulose. Colloids Surf. Physicochem.

Eng. Asp. 142, 75–82 (1998).

44. Araki, J., Wada, M., Kuga, S. & Okano, T. Birefringent Glassy Phase of a Cellulose

Microcrystal Suspension. Langmuir 16, 2413–2415 (2000).

45. Dong, X. M., Kimura, T., Revol, J.-F. & Gray, D. G. Effects of Ionic Strength on the

Isotropic−Chiral Nematic Phase Transition of Suspensions of Cellulose Crystallites.

Langmuir 12, 2076–2082 (1996).

112

46. Gama, M., Gatenholm, P. & Klemm, D. Bacterial NanoCellulose: A Sophisticated

Multifunctional Material. (CRC Press, 2012).

47. Nimeskern, L. et al. Mechanical evaluation of bacterial nanocellulose as an implant

material for ear cartilage replacement. J. Mech. Behav. Biomed. Mater. 22, 12–21 (2013).

48. Klemm, D. et al. Nanocelluloses: A New Family of Nature-Based Materials. Angew.

Chem. Int. Ed. 50, 5438–5466 (2011).

49. Wågberg, L. et al. The Build-Up of Polyelectrolyte Multilayers of Microfibrillated

Cellulose and Cationic Polyelectrolytes. Langmuir 24, 784–795 (2008).

50. Fleming, K., Gray, D., Prasannan, S. & Matthews, S. Cellulose Crystallites:  A New and

Robust Liquid Crystalline Medium for the Measurement of Residual Dipolar Couplings. J.

Am. Chem. Soc. 122, 5224–5225 (2000).

51. Alemdar, A. & Sain, M. Isolation and characterization of nanofibers from agricultural

residues – Wheat straw and soy hulls. Bioresour. Technol. 99, 1664–1671 (2008).

52. Dufresne, A., Cavaille, J.-Y. & Vignon, M. R. Mechanical behavior of sheets prepared

from sugar beet cellulose microfibrils. J. Appl. Polym. Sci. 64, 1185–1194 (1997).

53. Roux, J. C. & Mayade, T. L. Modeling of the particle breakage kinetics in the wet mills

for the paper industry. Powder Technol. 105, 237–242 (1999).

54. Turbak, A. F., Snyder, F. W. & Sandberg, K. R. Microfibrillated Cellulose, a New

Cellulose Product: Properties, Uses, and Commercial Potential. J Appl Polym Sci Appl

Polym Symp U. S. 37, (1983).

55. Nakagaito, A. N. & Yano, H. The effect of morphological changes from pulp fiber

towards nano-scale fibrillated cellulose on the mechanical properties of high-strength

plant fiber based composites. Appl. Phys. A 78, 547–552 (2003).

56. Hamad, W. Y. Some microrheological aspects of wood-pulp fibres subjected to fatigue

loading. Cellulose 4, 51–56

57. Eriksen, O., Syverud, K. & Gregersen, O. The use of microfibrillated cellulose produced

from kraft pulp as strength enhancer in TMP paper. Nord. Pulp Pap. Res. J. 23, 299–304

(2008).

58. Lindström, T. & Ankerfors, M. Seventh International Paper and Coating Chemistry

Symposium. Hamilt. Can. June 10–12 (2009).

59. Pääkkö, M. et al. Enzymatic Hydrolysis Combined with Mechanical Shearing and High-

Pressure Homogenization for Nanoscale Cellulose Fibrils and Strong Gels.

Biomacromolecules 8, 1934–1941 (2007).

60. Isogai, A., Saito, T. & Fukuzumi, H. TEMPO-oxidized cellulose nanofibers. Nanoscale 3,

71–85 (2011).

61. Wågberg, L., Winter, L., Ödberg, L. & Lindström, T. On the charge stoichiometry upon

adsorption of a cationic polyelectrolyte on cellulosic materials. Colloids Surf. 27, 163–

173 (1987).

62. Henriksson, M., Henriksson, G., Berglund, L. A. & Lindström, T. An environmentally

friendly method for enzyme-assisted preparation of microfibrillated cellulose (MFC)

nanofibers. Eur. Polym. J. 43, 3434–3441 (2007).

63. Eichhorn, S. J. et al. Review: current international research into cellulose nanofibres and

nanocomposites. J. Mater. Sci. 45, 1 (2009).

64. Horvath, A. E. & Lindström, T. The influence of colloidal interactions on fiber network

strength. J. Colloid Interface Sci. 309, 511–517 (2007).

65. Nechyporchuk, O., Belgacem, M. N. & Pignon, F. Rheological properties of micro-

/nanofibrillated cellulose suspensions: Wall-slip and shear banding phenomena.

Carbohydr. Polym. 112, 432–439 (2014).

66. Jia, X. et al. Rheological properties of an amorphous cellulose suspension. Food

Hydrocoll. 39, 27–33 (2014).

113

67. Iotti, M., Gregersen, Ø. W., Moe, S. & Lenes, M. Rheological Studies of Microfibrillar

Cellulose Water Dispersions. J. Polym. Environ. 19, 137–145 (2010).

68. Agoda-Tandjawa, G. et al. Rheological characterization of microfibrillated cellulose

suspensions after freezing. Carbohydr. Polym. 80, 677–686 (2010).

69. Karppinen, A. et al. Flocculation of microfibrillated cellulose in shear flow. Cellulose 19,

1807–1819 (2012).

70. Karppinen, A., Vesterinen, A.-H., Saarinen, T., Pietikäinen, P. & Seppälä, J. Effect of

cationic polymethacrylates on the rheology and flocculation of microfibrillated. Cellulose

18, 1381–1390 (2011).

71. Barnes, H. A. Thixotropy—a review. J. Non-Newton. Fluid Mech. 70, 1–33 (1997).

72. Rees, A. et al. 3D bioprinting of carboxymethylated-periodate oxidized nanocellulose

constructs for wound dressing applications. BioMed Res. Int. (2014).

73. Kumar, V. et al. Comparison of nano- and microfibrillated cellulose films. Cellulose 21,

3443–3456 (2014).

74. Sehaqui, H., Liu, A., Zhou, Q. & Berglund, L. A. Fast Preparation Procedure for Large,

Flat Cellulose and Cellulose/Inorganic Nanopaper Structures. Biomacromolecules 11,

2195–2198 (2010).

75. Henriksson, M., Berglund, L. A., Isaksson, P., Lindström, T. & Nishino, T. Cellulose

Nanopaper Structures of High Toughness. Biomacromolecules 9, 1579–1585 (2008).

76. Sehaqui, H. Nanofiber networks, aerogels and biocomposites based on nanofibrillated

cellulose from wood. (2011).

77. Spence, K. L., Venditti, R. A., Rojas, O. J., Habibi, Y. & Pawlak, J. J. A comparative

study of energy consumption and physical properties of microfibrillated cellulose

produced by different processing methods. Cellulose 18, 1097–1111 (2011).

78. Baez, C., Considine, J. & Rowlands, R. Influence of drying restraint on physical and

mechanical properties of nanofibrillated cellulose films. Cellulose 21, 347–356 (2014).

79. Jentzen, C. A. The effect of stress applied during drying on some of the properties of

individual pulp fibers. Tappi 47, 412–418 (1964).

80. Moon, R. J., Martini, A., Nairn, J., Simonsen, J. & Youngblood, J. Cellulose

nanomaterials review: structure, properties and nanocomposites. Chem. Soc. Rev. 40,

3941–3994 (2011).

81. Virtanen, S., Vartianen, J., Setälä, H., Tammelin, T. & Vuoti, S. Modified nanofibrillated

cellulose–polyvinyl alcohol films with improved mechanical performance. RSC Adv. 4,

11343 (2014).

82. Klemm, D. et al. Nanocelluloses as Innovative Polymers in Research and Application. in

Polysaccharides II (ed. Klemm, D.) 49–96 (Springer Berlin Heidelberg, 2006).

doi:10.1007/12_097

83. Jabbour, L. et al. Microfibrillated cellulose–graphite nanocomposites for highly flexible

paper-like Li-ion battery electrodes. J. Mater. Chem. 20, 7344 (2010).

84. Chiappone, A. et al. Microfibrillated cellulose as reinforcement for Li-ion battery

polymer electrolytes with excellent mechanical stability. J. Power Sources 196, 10280–

10288 (2011).

85. Kenneth, A. E. B. Lignosulfonate/urea binder for particulate composites. (1987).

86. Browning, W. C. Lignosulfonate Challenge. in Applied Polymer Symposia 28, 109–124

(DTIC Document, 1975).

87. Gardon, J. L. & Mason, S. G. Physicochemical Studies of Ligninsulphonates: I.

Preparation and Properties of Fractionated Samples. Can. J. Chem. 33, 1477–1490 (1955).

88. Heikkila, H. Production of pure sugars and lignosulfonates from sulfite spent liquor.

89. Fredheim, G. E. & Christensen, B. E. Polyelectrolyte Complexes:  Interactions between

Lignosulfonate and Chitosan. Biomacromolecules 4, 232–239 (2003).

114

90. Lemes, A. P., Soto-Oviedo, M. A., Waldman, W. R., Innocentini-Mei, L. H. & Durán, N.

Effect of Lignosulfonate on the Thermal and Morphological Behavior of Poly(3-

hydroxybutyrate-co-3-hydroxyvalerate). J. Polym. Environ. 18, 250–259 (2010).

91. Kauppi, A., Andersson, K. M. & Bergström, L. Probing the effect of superplasticizer

adsorption on the surface forces using the colloidal probe AFM technique. Cem. Concr.

Res. 35, 133–140 (2005).

92. Shulga, G., Rekner, F. & Varslavan, J. SW—Soil and Water. J. Agric. Eng. Res. 78, 309–

316 (2001).

93. Huang, J., Zhang, L. & Chen, F. Effects of lignin as a filler on properties of soy protein

plastics. I. Lignosulfonate. J. Appl. Polym. Sci. 88, 3284–3290 (2003).

94. Vainio, U., Lauten, R. A. & Serimaa, R. Small-Angle X-ray Scattering and Rheological

Characterization of Aqueous Lignosulfonate Solutions. Langmuir 24, 7735–7743 (2008).

95. Barnes, H. A. & Hutton, J. F. An Introduction to Rheology. (Elsevier, 1989).

96. Marcus, Y. The sizes of molecules—revisited. J. Phys. Org. Chem. 16, 398–408 (2003).

97. Werff, J. C. van der & Kruif, C. G. de. Hard‐sphere Colloidal Dispersions: The Scaling of

Rheological Properties with Particle Size, Volume Fraction, and Shear Rate. J. Rheol.

1978-Present 33, 421–454 (1989).

98. Luckham, P. F. & Ukeje, M. A. Effect of Particle Size Distribution on the Rheology of

Dispersed Systems. J. Colloid Interface Sci. 220, 347–356 (1999).

99. Branca, C. & Blasi, C. D. A lumped kinetic model for banana peel combustion.

Thermochim. Acta 614, 68–75 (2015).

100. Koufopanos, C. A., Lucchesi, A. & Maschio, G. Kinetic modelling of the pyrolysis of

biomass and biomass components. Can. J. Chem. Eng. 67, 75–84 (1989).

101. Branca, C., Di Blasi, C. & Horacek, H. Analysis of the Combustion Kinetics and

Thermal Behavior of an Intumescent System. Ind. Eng. Chem. Res. 41, 2107–2114 (2002).

102. White, J. E., Catallo, W. J. & Legendre, B. L. Biomass pyrolysis kinetics: A

comparative critical review with relevant agricultural residue case studies. J. Anal. Appl.

Pyrolysis 91, 1–33 (2011).

103. Galwey, A. K. & Brown, M. E. Application of the Arrhenius equation to solid state

kinetics: can this be justified? Thermochim. Acta 386, 91–98 (2002).

104. Atkins, P. W. Physical Chemistry. 6th. (Oxford University Press, 1998).

105. Garn, P. D. Kinetics of thermal decomposition of the solid state: II. Delimiting the

homogeneous-reaction model. Thermochim. Acta 160, 135–145 (1990).

106. Amutio, M. et al. Kinetic study of lignocellulosic biomass oxidative pyrolysis. Fuel 95,

305–311 (2012).

107. Anca-Couce, A., Berger, A. & Zobel, N. How to determine consistent biomass

pyrolysis kinetics in a parallel reaction scheme. Fuel 123, 230–240 (2014).

108. Antal, M. J. J. & Varhegyi, G. Cellulose pyrolysis kinetics: the current state of

knowledge. Ind. Eng. Chem. Res. 34, 703–717 (1995).

109. Raghavan, V., Cohen, M. & Hannay, N. B. Treatise on Solid State Chemistry. Vol

5Plenum N. Y. 96 (1975).

110. Le Claire, A. D. & Hannay, N. B. Treatise on Solid State Chemistry. Vol 4 1 (1975).

111. Vyazovkin, S. & Wight, C. A. Isothermal and non-isothermal kinetics of thermally

stimulated reactions of solids. Int. Rev. Phys. Chem. 17, 407–433 (1998).

112. Bruzs, B. Velocity of Thermal Decomposition of Carbonates. J. Phys. Chem. 30, 680–

693 (1925).

113. Hüttig, G. F., Meller, A. & Lehmann, E. Die geschwindigkeit des zerfalls von

zinkcarbonat in zinkoxyd und kohlendioxyd. Z Phys Chem B 19, 1–21 (1932).

114. Koga, N. & Šesták, J. Kinetic compensation effect as a mathematical consequence of

the exponential rate constant. Thermochim. Acta 182, 201–208 (1991).

115

115. Kujirai, T. & Akahira, T. Effect of temperature on the deterioration of fibrous

insulating materials. Sci Pap Inst Phys Chem Res 2, 223–52 (1925).

116. Flynn, J. H. The ‘Temperature Integral’ — Its use and abuse. Thermochim. Acta 300,

83–92 (1997).

117. Flynn, J. H. & Wall, L. A. General treatment of the thermogravimetry of polymers. J

Res Nat Bur Stand 70, 487–523 (1966).

118. Ozawa, T. A new method of analyzing thermogravimetric data. Bull. Chem. Soc. Jpn.

38, 1881–1886 (1965).

119. Doyle, C. D. Estimating isothermal life from thermogravimetric data. J. Appl. Polym.

Sci. 6, 639–642 (1962).

120. Vyazovkin, S. & Dollimore, D. Linear and Nonlinear Procedures in Isoconversional

Computations of the Activation Energy of Nonisothermal Reactions in Solids. J. Chem.

Inf. Comput. Sci. 36, 42–45 (1996).

121. Cai, J. M. & Bi, L. S. Kinetic analysis of wheat straw pyrolysis using isoconversional

methods. J. Therm. Anal. Calorim. 98, 325–330 (2009).

122. Cagnon, B., Py, X., Guillot, A., Stoeckli, F. & Chambat, G. Contributions of

hemicellulose, cellulose and lignin to the mass and the porous properties of chars and

steam activated carbons from various lignocellulosic precursors. Bioresour. Technol. 100,

292–298 (2009).

123. Yang, H., Yan, R., Chen, H., Lee, D. H. & Zheng, C. Characteristics of hemicellulose,

cellulose and lignin pyrolysis. Fuel 86, 1781–1788 (2007).

124. Yang, H. et al. In-Depth Investigation of Biomass Pyrolysis Based on Three Major

Components:  Hemicellulose, Cellulose and Lignin. Energy Fuels 20, 388–393 (2006).

125. Worasuwannarak, N., Sonobe, T. & Tanthapanichakoon, W. Pyrolysis behaviors of

rice straw, rice husk, and corncob by TG-MS technique. J. Anal. Appl. Pyrolysis 78, 265–

271 (2007).

126. Stefanidis, S. D. et al. A study of lignocellulosic biomass pyrolysis via the pyrolysis of

cellulose, hemicellulose and lignin. J. Anal. Appl. Pyrolysis 105, 143–150 (2014).

127. Kim, S. & Eom, Y. Estimation of kinetic triplet of cellulose pyrolysis reaction from

isothermal kinetic results. Korean J. Chem. Eng. 23, 409–414 (2006).

128. Müller-Hagedorn, M., Bockhorn, H., Krebs, L. & Müller, U. A comparative kinetic

study on the pyrolysis of three different wood species. J. Anal. Appl. Pyrolysis 68–69,

231–249 (2003).

129. Jakab, E., Faix, O., Till, F. & Székely, T. Thermogravimetry/Mass Spectrometry of

Various Lignosulfonates as well as of a Kraft and Acetosolv Lignin. Holzforsch. - Int. J.

Biol. Chem. Phys. Technol. Wood 45, 355–360 (2009).

130. Jakab, E., Faix, O., Till, F. & Székely, T. Proceedings of the 10th International

Conference on Fundamental Aspects, Processes and Applications of PyrolysisThe effect

of cations on the thermal decomposition of lignins. J. Anal. Appl. Pyrolysis 25, 185–194

(1993).

131. Li, B., Lv, W., Zhang, Q., Wang, T. & Ma, L. Pyrolysis and catalytic pyrolysis of

industrial lignins by TG-FTIR: Kinetics and products. J. Anal. Appl. Pyrolysis 108, 295–

300 (2014).

132. Liu, Q., Zhong, Z., Wang, S. & Luo, Z. Interactions of biomass components during

pyrolysis: A TG-FTIR study. J. Anal. Appl. Pyrolysis 90, 213–218 (2011).

133. Wang, S., Guo, X., Wang, K. & Luo, Z. Influence of the interaction of components on

the pyrolysis behavior of biomass. J. Anal. Appl. Pyrolysis 91, 183–189 (2011).

134. Hosoya, T., Kawamoto, H. & Saka, S. Cellulose–hemicellulose and cellulose–lignin

interactions in wood pyrolysis at gasification temperature. J. Anal. Appl. Pyrolysis 80,

118–125 (2007).

116

135. Rao, T. R. & Sharma, A. Pyrolysis rates of biomass materials. Energy 23, 973–978

(1998).

136. Manyà, J. J., Velo, E. & Puigjaner, L. Kinetics of Biomass Pyrolysis:  a Reformulated

Three-Parallel-Reactions Model. Ind. Eng. Chem. Res. 42, 434–441 (2003).

137. Orfao, J. J. M., Antunes, F. J. A. & Figueiredo, J. L. Pyrolysis kinetics of

lignocellulosic materials—three independent reactions model. Fuel 78, 349–358 (1999).

138. Raveendran, K., Ganesh, A. & Khilar, K. C. Influence of mineral matter on biomass

pyrolysis characteristics. Fuel 74, 1812–1822 (1995).

139. Nassar, M. M. Thermal Analysis Kinetics of Bagasse and Rice Straw. Energy Sources

20, 831–837 (1998).

140. Varhegyi, G., Antal, M. J., Szekely, T., Till, F. & Jakab, E. Simultaneous

thermogravimetric-mass spectrometric studies of the thermal decomposition of

biopolymers. 1. Avicel cellulose in the presence and absence of catalysts. Energy Fuels 2,

267–272 (1988).

141. Brunner, P. H. & Roberts, P. V. The significance of heating rate on char yield and char

properties in the pyrolysis of cellulose. Carbon 18, 217–224 (1980).

142. Ahmed, I. I. & Gupta, A. K. Pyrolysis and gasification of food waste: Syngas

characteristics and char gasification kinetics. Appl. Energy 87, 101–108 (2010).

143. Dumanlı, A. G. & Windle, A. H. Carbon fibres from cellulosic precursors: a review. J.

Mater. Sci. 47, 4236–4250 (2012).

144. Gao, M., Ling, B., Yang, S. & Zhao, M. Flame retardance of wood treated with

guanidine compounds characterized by thermal degradation behavior. J. Anal. Appl.

Pyrolysis 73, 151–156 (2005).

145. Zeng, F., Pan, D. & Pan, N. Choosing the Impregnants by Thermogravimetric

Analysis for Preparing Rayon-Based Carbon Fibers. J. Inorg. Organomet. Polym. Mater.

15, 261–267 (2005).

146. Lysenko, A. A., Piskunova, I. A. & Astashkina, O. V. Effect of Inorganic and

Organosilicon Additives on Pyrolysis of Hydrated Cellulose Fibre Materials. Fibre Chem.

35, 189–192 (2003).

147. Kim, D.-Y., Nishiyama, Y., Wada, M. & Kuga, S. High-yield Carbonization of

Cellulose by Sulfuric Acid Impregnation. Cellulose 8, 29–33 (2001).

148. Ishida, O., Kim, D.-Y., Kuga, S., Nishiyama, Y. & Brown, R. M. Microfibrillar carbon

from native cellulose. Cellulose 11, 475–480 (2004).

149. Sevilla, M. & Fuertes, A. B. Graphitic carbon nanostructures from cellulose. Chem.

Phys. Lett. 490, 63–68 (2010).

150. McDonald-Wharry, J. S., Manley-Harris, M. & Pickering, K. L. Reviewing,

Combining, and Updating the Models for the Nanostructure of Non-Graphitizing Carbons

Produced from Oxygen-Containing Precursors. Energy Fuels (2016).

doi:10.1021/acs.energyfuels.6b00917

151. Radovic, L. R. Chemistry & Physics of Carbon. (CRC Press, 2000).

152. Guerrero, M., Ruiz, M. P., Millera, Á., Alzueta, M. U. & Bilbao, R. Characterization

of Biomass Chars Formed under Different Devolatilization Conditions: Differences

between Rice Husk and Eucalyptus. Energy Fuels 22, 1275–1284 (2008).

153. Xin, S. et al. Chemical structure evolution of char during the pyrolysis of cellulose. J.

Anal. Appl. Pyrolysis 116, 263–271 (2015).

154. Oschatz, M. et al. Nanostructure characterization of carbide-derived carbons by

morphological analysis of transmission electron microscopy images combined with

physisorption and Raman spectroscopy. Carbon 105, 314–322 (2016).

117

155. Pré, P. et al. A new approach to characterize the nanostructure of activated carbons

from mathematical morphology applied to high resolution transmission electron

microscopy images. Carbon 52, 239–258 (2013).

156. Kercher, A. K. & Nagle, D. C. Evaluation of carbonized medium-density fiberboard

for electrical applications. Carbon 40, 1321–1330 (2002).

157. Shao, Y., Guizani, C., Grosseau, P., Chaussy, D. & Beneventi, D. Thermal

characterization and kinetic analysis of microfibrillated cellulose/lignosulfonate blends. J.

Anal. Appl. Pyrolysis 124, 25–34 (2017).

158. Rhim, Y.-R. et al. Changes in the thermophysical properties of microcrystalline

cellulose as function of carbonization temperature. Carbon 48, 31–40 (2010).

159. Parfen’eva, L. S. et al. Thermal conductivity of high-porosity biocarbon preforms of

beech wood. Phys. Solid State 52, 1115–1122 (2010).

160. Tzeng, S.-S. & Chr, Y.-G. Evolution of microstructure and properties of phenolic

resin-based carbon/carbon composites during pyrolysis. Mater. Chem. Phys. 73, 162–169

(2002).

161. Smits, F. . Measurement of sheet resistivities with the four-point probe. Bell Syst. Tech.

J. 37, 711–718 (1958).

162. Tsai, W.-T., Liu, S.-C., Chen, H.-R., Chang, Y.-M. & Tsai, Y.-L. Textural and

chemical properties of swine-manure-derived biochar pertinent to its potential use as a

soil amendment. Chemosphere 89, 198–203 (2012).

163. Yamauchi, S. & Kurimoto, Y. Raman spectroscopic study on pyrolyzed wood and

bark of Japanese cedar: temperature dependence of Raman parameters. J. Wood Sci. 49,

235–240

164. Li, X., Hayashi, J. & Li, C.-Z. FT-Raman spectroscopic study of the evolution of char

structure during the pyrolysis of a Victorian brown coal. Fuel 85, 1700–1707 (2006).

165. Li, X., Hayashi, J. & Li, C.-Z. Volatilisation and catalytic effects of alkali and alkaline

earth metallic species during the pyrolysis and gasification of Victorian brown coal. Part

VII. Raman spectroscopic study on the changes in char structure during the catalytic

gasification in air. Fuel 85, 1509–1517 (2006).

166. Zhao, Y., Feng, D., Zhang, Y., Huang, Y. & Sun, S. Effect of pyrolysis temperature on

char structure and chemical speciation of alkali and alkaline earth metallic species in

biochar. Fuel Process. Technol. 141, Part 1, 54–60 (2016).

167. McDonald-Wharry, J., Manley-Harris, M. & Pickering, K. Carbonisation of biomass-

derived chars and the thermal reduction of a graphene oxide sample studied using Raman

spectroscopy. Carbon 59, 383–405 (2013).

168. Kim, C. et al. Raman spectroscopic evaluation of polyacrylonitrile-based carbon

nanofibers prepared by electrospinning. J. Raman Spectrosc. 35, 928–933 (2004).

169. Manoj, B. & Kunjomana, A. G. Study of stacking structure of amorphous carbon by

X-ray diffraction technique. Int J Electrochem Sci 7, 3127–3134 (2012).

170. Takagi, H., Maruyama, K., Yoshizawa, N., Yamada, Y. & Sato, Y. XRD analysis of

carbon stacking structure in coal during heat treatment. Fuel 83, 2427–2433 (2004).

171. Lewis, J. A., Smay, J. E., Stuecker, J. & Cesarano, J. Direct Ink Writing of Three-

Dimensional Ceramic Structures. J. Am. Ceram. Soc. 89, 3599–3609 (2006).

172. Sorvari, A. et al. Modifying the flocculation of microfibrillated cellulose suspensions

by soluble polysaccharides under conditions unfavorable to adsorption. Carbohydr. Polym.

106, 283–292 (2014).

173. Charani, P. R., Dehghani-Firouzabadi, M., Afra, E. & Shakeri, A. Rheological

characterization of high concentrated MFC gel from kenaf unbleached pulp. Cellulose 20,

727–740 (2013).

118

174. Mewis, J. & Wagner, N. J. Thixotropy. Adv. Colloid Interface Sci. 147–148, 214–227

(2009).

175. Zhao, R. H. & Kerekes, R. J. The effect of suspending liquid viscosity on fiber

flocculation. Tappi J. 76, 183–188 (1993).

176. Deville, S., Saiz, E. & Tomsia, A. P. Ice-templated porous alumina structures. Acta

Mater. 55, 1965–1974 (2007).

177. Deville, S., Saiz, E. & Tomsia, A. P. Freeze casting of hydroxyapatite scaffolds for

bone tissue engineering. Biomaterials 27, 5480–5489 (2006).

178. Teng, N.-Y., Dallmeyer, I. & Kadla, J. F. Incorporation of Multiwalled Carbon

Nanotubes into Electrospun Softwood Kraft Lignin-Based Fibers. J. Wood Chem. Technol.

33, 299–316 (2013).

179. Snowdon, M. R., Mohanty, A. K. & Misra, M. A Study of Carbonized Lignin as an

Alternative to Carbon Black. ACS Sustain. Chem. Eng. 2, 1257–1263 (2014).

180. Deraman, M., Zakaria, S., Omar, R. & Aziz, A. A. Electrical Conductivity of Carbon

Pellets from Mixtures of Pyropolymer from Oil Palm Bunch and Cotton Cellulose. Jpn. J.

Appl. Phys. 39, L1236 (2000).

181. Lu, X., Nilsson, O. & Pekala, J. F. R. W. Thermal and electrical conductivity of

monolithic carbon aerogels. J. Appl. Phys. 73, 581–584 (1993).

182. Sánchez-González, J., Macías-García, A., Alexandre-Franco, M. F. & Gómez-Serrano,

V. Electrical conductivity of carbon blacks under compression. Carbon 43, 741–747

(2005).

183. Henriksson, M. & Berglund, L. A. Structure and properties of cellulose nanocomposite

films containing melamine formaldehyde. J. Appl. Polym. Sci. 106, 2817–2824 (2007).

184. Quiévy, N. et al. Influence of homogenization and drying on the thermal stability of

microfibrillated cellulose. Polym. Degrad. Stab. 95, 306–314 (2010).

185. Aulin, C., Gällstedt, M. & Lindström, T. Oxygen and oil barrier properties of

microfibrillated cellulose films and coatings. Cellulose 17, 559–574 (2010).

186. Poletto, M., Pistor, V., Zeni, M. & Zattera, A. J. Crystalline properties and

decomposition kinetics of cellulose fibers in wood pulp obtained by two pulping

processes. Polym. Degrad. Stab. 96, 679–685 (2011).

187. Hokkanen, S., Bhatnagar, A., Repo, E., Lou, S. & Sillanpää, M. Calcium

hydroxyapatite microfibrillated cellulose composite as a potential adsorbent for the

removal of Cr(VI) from aqueous solution. Chem. Eng. J. 283, 445–452 (2016).

188. Ge, Y., Li, Z., Pang, Y. & Qiu, X. Influence of molecular mass of lignosulfonates on

the resulting surface charges of solid particles. Int. J. Biol. Macromol. 52, 300–304 (2013).

189. Jakab, E., Faix, O., Till, F. & Székely, T. The effect of cations on the thermal

decomposition of lignins. J. Anal. Appl. Pyrolysis 25, 185–194 (1993).

190. Wang, J. et al. Catalytic effects of six inorganic compounds on pyrolysis of three

kinds of biomass. Thermochim. Acta 444, 110–114 (2006).

191. Amen-Chen, C., Pakdel, H. & Roy, C. Production of monomeric phenols by

thermochemical conversion of biomass: a review. Bioresour. Technol. 79, 277–299

(2001).

192. Várhegyi, G., Till, F. & Székely, T. Software for a mass spectrometer-thermobalance

system. Thermochim. Acta 102, 115–124 (1986).

193. van Loon, W. M. G. M., Boon, J. J. & de Groot, B. Qualitative analysis of

chlorolignins and lignosulphonates in pulp mill effluents entering the river Rhine using

pyrolysis—mass spectrometry and pyrolysis—gas chromatography/mass spectrometry. J.

Anal. Appl. Pyrolysis 20, 275–302 (1991).

194. van Loon, W. M. G. M., Boon, J. J. & de Groot, B. Quantitative analysis of sulfonic

acid groups in macromolecular lignosulfonic acids and aquatic humic substances by

119

temperature-resolved pyrolysis-mass spectrometry. Environ. Sci. Technol. 27, 2387–2396

(1993).

195. Grønli, M. G., Várhegyi, G. & Di Blasi, C. Thermogravimetric Analysis and

Devolatilization Kinetics of Wood. Ind. Eng. Chem. Res. 41, 4201–4208 (2002).

196. Várhegyi, G., Antal Jr., M. J., Jakab, E. & Szabó, P. Kinetic modeling of biomass

pyrolysis. J. Anal. Appl. Pyrolysis 42, 73–87 (1997).

197. McDonald-Wharry, J., Manley-Harris, M. & Pickering, K. A comparison of the

charring and carbonisation of oxygen-rich precursors with the thermal reduction of

graphene oxide. Philos. Mag. 95, 4054–4077 (2015).

198. Lin-Vien, D., Colthup, N. B., Fateley, W. G. & Grasselli, J. G. The Handbook of

Infrared and Raman Characteristic Frequencies of Organic Molecules. (Elsevier, 1991).

199. Keiluweit, M., Nico, P. S., Johnson, M. G. & Kleber, M. Dynamic Molecular

Structure of Plant Biomass-Derived Black Carbon (Biochar). Environ. Sci. Technol. 44,

1247–1253 (2010).

200. Bohn Lima, R. et al. Direct lignin fuel cell for power generation. RSC Adv. 3, 5083–

5089 (2013).

201. Mochidzuki, K. et al. Electrical and Physical Properties of Carbonized Charcoals. Ind.

Eng. Chem. Res. 42, 5140–5151 (2003).

202. Guo, Y. & Bustin, R. M. FTIR spectroscopy and reflectance of modern charcoals and

fungal decayed woods: implications for studies of inertinite in coals. Int. J. Coal Geol. 37,

29–53 (1998).

203. McKee, D. W., Spiro, C. L. & Lamby, E. J. The inhibition of graphite oxidation by

phosphorus additives. Carbon 22, 285–290 (1984).

204. Mapelli, C., Castiglioni, C., Meroni, E. & Zerbi, G. Graphite and graphitic compounds:

vibrational spectra from oligomers to real materials. J. Mol. Struct. 480–481, 615–620

(1999).

205. Brewer, C. E. et al. New approaches to measuring biochar density and porosity.

Biomass Bioenergy 66, 176–185 (2014).

206. Lehmann, J. Biochar for Environmental Management: Science and Technology.

(Earthscan, 2012).

207. Kercher, A. K. & Nagle, D. C. Microstructural evolution during charcoal

carbonization by X-ray diffraction analysis. Carbon 41, 15–27 (2003).

208. Brown, R. A., Kercher, A. K., Nguyen, T. H., Nagle, D. C. & Ball, W. P. Production

and characterization of synthetic wood chars for use as surrogates for natural sorbents.

Org. Geochem. 37, 321–333 (2006).

209. Ferrari, A. C. & Robertson, J. Raman spectroscopy of amorphous, nanostructured,

diamond–like carbon, and nanodiamond. Philos. Trans. R. Soc. Lond. Math. Phys. Eng.

Sci. 362, 2477–2512 (2004).

210. shimodaira, N. Raman spectroscopic investigations of activated carbon materials. J.

Appl. Phys. 92, 902–909 (2002).

211. Cuesta, A., Dhamelincourt, P., Laureyns, J., Martínez-Alonso, A. & Tascón, J. M. D.

Raman microprobe studies on carbon materials. Carbon 32, 1523–1532 (1994).

212. Zickler, G. A., Smarsly, B., Gierlinger, N., Peterlik, H. & Paris, O. A reconsideration

of the relationship between the crystallite size La of carbons determined by X-ray

diffraction and Raman spectroscopy. Carbon 44, 3239–3246 (2006).

213. Mattevi, C. et al. Evolution of Electrical, Chemical, and Structural Properties of

Transparent and Conducting Chemically Derived Graphene Thin Films. Adv. Funct.

Mater. 19, 2577–2583 (2009).

120

214. Liu, X.-Q., Ding, H.-S., Wang, Y.-Y., Liu, W.-J. & Jiang, H. Pyrolytic Temperature

Dependent and Ash Catalyzed Formation of Sludge Char with Ultra-High Adsorption to

1-Naphthol. Environ. Sci. Technol. 50, 2602–2609 (2016).

215. Guizani, C., Haddad, K., Limousy, L. & Jeguirim, M. New insights on the structural

evolution of biomass char upon pyrolysis as revealed by the Raman spectroscopy and

elemental analysis. Carbon doi:10.1016/j.carbon.2017.04.078

216. Okumura, Y., Hanaoka, T. & Sakanishi, K. Effect of pyrolysis conditions on

gasification reactivity of woody biomass-derived char. Proc. Combust. Inst. 32, 2013–

2020 (2009).

217. Paris, O., Zollfrank, C. & Zickler, G. A. Decomposition and carbonisation of wood

biopolymers—a microstructural study of softwood pyrolysis. Carbon 43, 53–66 (2005).

218. Frank, O. et al. Compression Behavior of Single-Layer Graphenes. ACS Nano 4,

3131–3138 (2010).

219. Huang, M. et al. Phonon softening and crystallographic orientation of strained

graphene studied by Raman spectroscopy. Proc. Natl. Acad. Sci. 106, 7304–7308 (2009).

220. del Corro, E., de la Roza, A. O., Taravillo, M. & Baonza, V. G. Raman modes and

Grüneisen parameters of graphite under compressive biaxial stress. Carbon 50, 4600–

4606 (2012).

221. Harris, P. J. F., Liu, Z. & Suenaga, K. Imaging the atomic structure of activated carbon.

J. Phys. Condens. Matter 20, 362201 (2008).

222. Liu, T., Luo, R., Qiao, W., Yoon, S.-H. & Mochida, I. Microstructure of carbon

derived from mangrove charcoal and its application in Li-ion batteries. Electrochimica

Acta 55, 1696–1700 (2010).

223. Aso, H., Matsuoka, K., Sharma, A. & Tomita, A. Structural analysis of PVC and PFA

carbons prepared at 500–1000 °C based on elemental composition, XRD, and HRTEM.

Carbon 42, 2963–2973 (2004).

224. Li, Z. Q., Lu, C. J., Xia, Z. P., Zhou, Y. & Luo, Z. X-ray diffraction patterns of

graphite and turbostratic carbon. Carbon 45, 1686–1695 (2007).

225. Sakintuna, B., Yürüm, Y. & Çetinkaya, S. Evolution of Carbon Microstructures during

the Pyrolysis of Turkish Elbistan Lignite in the Temperature Range 700−1000 °C. Energy

Fuels 18, 883–888 (2004).

226. Ashby, M. F. et al. Metal Foams: A Design Guide. (Elsevier, 2000).

227. Roig, F., Dantras, E., Dandurand, J. & Lacabanne, C. Influence of hydrogen bonds on

glass transition and dielectric relaxations of cellulose. J. Phys. Appl. Phys. 44, 045403

(2011).

228. Chen, W. et al. Individualization of cellulose nanofibers from wood using high-

intensity ultrasonication combined with chemical pretreatments. Carbohydr. Polym. 83,

1804–1811 (2011).

229. Jonoobi, M., Harun, J., Mishra, M. & Oksman, K. Chemical composition, crystallinity

and thermal degradation of bleached and unbleached kenaf bast (Hibiscus cannabinus)

pulp and nanofiber. BioResources 4, 626–639 (2009).

230. Hokkanen, S. et al. Adsorption of Ni(II), Cu(II) and Cd(II) from aqueous solutions by

amino modified nanostructured microfibrillated cellulose. Cellulose 21, 1471–1487

(2014).

231. Jonoobi, M., Khazaeian, A., Tahir, P. M., Azry, S. S. & Oksman, K. Characteristics of

cellulose nanofibers isolated from rubberwood and empty fruit bunches of oil palm using

chemo-mechanical process. Cellulose 18, 1085–1095 (2011).

232. Rosa, M. F. et al. Cellulose nanowhiskers from coconut husk fibers: Effect of

preparation conditions on their thermal and morphological behavior. Carbohydr. Polym.

81, 83–92 (2010).

121

233. Lu, H., Gui, Y., Zheng, L. & Liu, X. Morphological, crystalline, thermal and

physicochemical properties of cellulose nanocrystals obtained from sweet potato residue.

Food Res. Int. 50, 121–128 (2013).

234. Eyholzer, C. et al. Preparation and characterization of water-redispersible

nanofibrillated cellulose in powder form. Cellulose 17, 19–30 (2010).

235. Gibson, L. J. & Ashby, M. F. Cellular Solids: Structure and Properties. (Cambridge

University Press, 1999).

122

6. Appendices

123

6 Appendices

A: Elemental analysis data of the used NaLS Element Concentration (%)

C 40.651

O 44.628

Na 9.043

S 5.325

Others: Mg, Al, Si, Cl, K, Ca 0.353

B.Elemental analysis data of MFC Element Molar percentage (%)

C 33.05

H 0.44

N

O 66.51

(copyright Hokkanen et al.187)

124

7. Published/submitted

papers

125

7 Published/submitted papers [1] Y. Shao, D. Chaussy, P. Grosseau, D. Beneventi, Use of Microfibrillated

Cellulose/Lignosulfonate Blends as Carbon Precursors: Impact of Hydrogel

Rheology on 3D Printing, Ind. Eng. Chem. Res. (2015).

doi:10.1021/acs.iecr.5b02763.

[2] Y. Shao, C. Guizani, P. Grosseau, D. Chaussy, D. Beneventi, Thermal

characterization and kinetic analysis of microfibrillated cellulose/lignosulfonate

blends, J. Anal. Appl. Pyrolysis. 124 (2017) 25–34.

doi:10.1016/j.jaap.2017.03.001.

[3] Y. Shao, C. Guizani, P. Grosseau, D. Chaussy, D. Beneventi, Biocarbons from

microfibrillated cellulose/lignosulfonate precursors: a study of electrical

conductivity development during slow pyrolysis, Carbon, Under review

[4] Y. Shao, C. Guizani, P. Grosseau, D. Chaussy, D. Beneventi, Use of purely lignocellulosic materials and 3D printing for the development of functional host materials in electrodes of energy storage devices, Bioresource Technol., Submission

126

8. French abstract

127

8 French abstract Cette thèse a pour l’objectif d’élaborer, à partir de technologie additive 3D (extrusion de gel),

des structures conductrices (électriquement) et résistantes (mécaniquement) en utilisant

exclusivement des matériaux lignocellulosiques. Les matériaux sélectionnés sont des

microfibrilles de cellulose (MFC), du lignosulfonate de sodium (LS ou NaLS) et de la

cellulose en poudre (CP). Ces trois constituants peuvent être utilisés pour élaborer des

hydrogels aqueux compatibles avec l’impression 3D (extrusion). De plus, ce procédé

d’impression permet la mise en forme avec une excellente définition de structures à base de

précurseurs de carbone dont les géométries peuvent être adaptées aux différentes techniques

de caractérisation sélectionnées.

Cette thèse a été divisée en 4 parties distinctes

La première concerne l’étude des propriétés rhéologiques des hydrogels de mélanges

MFC/LS et leur compatibilité avec le procédé d’extrusion ainsi que la stabilité dimensionnelle

des structures 3D imprimées.

Le comportement rhéologique des hydrogels de MFC/LS est présenté sur la Figure 1. En

général, les hydrogels sont rhéofluidifiants, ce qui est très favorable pour ce type d’impression

3D. En effet, afin de pouvoir facilement extruder le gel au niveau de la buse d’impression, il

est nécessaire que la viscosité du fluide diminue lorsqu’il est sujet à un cisaillement.

Figure 1 Viscosité en fonction du taux de cisaillement des hydrogels de MFC/LS : a) Série à 2%

MFC, la figure insérée représente la contrainte en fonction du taux de cisaillement; b) Série à

1% MFC. Les lignes pointillées représentent les modèles obtenus à partir d’Équation 1et

Équation 2.

128

Équation 1 𝜎 = 𝜎𝑦 + 𝐾��𝑛 + 𝜂𝑠��

Équation 2 𝜂 = 𝜎𝑦

��+ 𝑛𝐾��𝑛−1 + 𝜂𝑠

Dans les conditions d’extrusion, le taux de cisaillement subit par l’hydrogel est d’environ 300

s-1

. Pour cette valeur de cisaillement la viscosité correspondante se doit d’être faible. Lorsque

l’hydrogel a été déposé sur le substrat et que le cisaillement a disparu, l’hydrogel doit

récupérer sa viscosité immédiatement afin d’avoir un comportant solide et éviter un étalement

de la matière. Autrement dit, il doit acquérir une viscosité assez élevée quand le cisaillement

qu’il subit est petit.

La thixotropie des hydrogels de MFC/LS est présentée sur la Figure 2. Le profil décrit un

comportement de restauration du réseau de matiere dans l’hydrogel apres un changement

brutal du taux de cisaillement (i.e. diminution de 1000 à 0.1s-1

). La figure 2 montre que pour

les hydrogels dont la concentration de LS est inférieure à 30%, la contrainte de cisaillement

présente une diminution rapide traduisant une réponse viscoélastique du réseau de MFC à une

brusque variation du taux de cisaillement. La constante de temps, obtenue à partir d’Équation

3 , augmente progressivement de 4 à 7 s quand la concentration de LS augmente de 0% à 30%,

ce qui signifie que la thixotropie du réseau de MFC diminue lorsque le taux de LS augmente.

Figure 2 La réponse de la contrainte de cisaillement après une chute brutale du taux de

cisaillement de 1000 à 0.1 s−1

pour la série de 2% MFC. Les lignes pointillées représentent les

résultats de fitting selon Équation 3.

Équation 3 𝜎(𝑡) = 𝜎0 + (𝜎∞ − 𝜎0) (1 − 𝑒−𝑡

𝜏) + 𝜂𝑠��

La diminution de la contrainte de cisaillement et son augmentation jusqu'à la stabilisation,

observée pour des concentrations de LS au-dessus de 40%, révèle une transition du

comportement viscoélastique vers un comportement thixotropique dominé par la composante

visqueuse. En raison de la viscosité accrue de la solution de LS, la reconstruction du réseau

129

MFC est retardée. L’augmentation de la concentration en LS jusqu'à 45% et 50% souligne la

réponse thixotropique avec un accroissement de la constante de temps de 0,7 à 14 et 27 s pour

des concentration respectives en LS de 40%, 45% et 50%.

Comme l'illustre la Figure 3, des structures cubiques avec une bonne définition spatiale (c'est-

à-dire des arêtes vives et sans étalement de la base du cube sur le substrat) ont été imprimées à

l'aide de l'hydrogel pur de MFC 2%. Les ajouts de LS ont entraîné une dégradation

progressive de la forme souhaitée. Lorsque la concentration en LS augmente de 0 à 30%, la

base du cube a tendance à s’étaler et sa hauteur à diminuer d'environ 9 à 2 mm,

respectivement. Selon les données expérimentales présentées en Figure 2, ce comportement a

été associé à une diminution de la viscosité (de 130 à 60 Pa s) à des taux de cisaillement

faibles. Avec 40% de LS, la suspension MFC/LS s'est complètement étalée sur le substrat et

l'épaisseur du film liquide final est inférieure à 1 mm. Cette tendance est attribuée à la

transition du comportement thixotropique viscoélastique à inélastique. En effet, après le

cisaillement dans l'aiguille de l'extrudeuse et la destruction du réseau MFC, la cinétique de

reconstruction du réseau est trop lente et la viscosité du fluide trop faible pour empêcher la

suspension MFC / LS de se répandre sur le substrat d'impression. Avec 50% de LS, les

structures cubiques sont facilement imprimées sans déformations majeures. Cependant, leurs

arêtes présentent des profils arrondis, indiquant un phénomène de fluage au sein de la

suspension MFC/LS.

Figure 3 : Photographies et principales caractéristiques des structures cubiques avant et après

séchage et pyrolyse.

LS: 0% 10% 30% 50%

Before drying

After air drying

Freeze drying necessary to hold the original shape

Not adapted for 3D printing

Lateral side Lateral sideTop side Top side

After freeze drying and pyrolysis (LS 4%)

After air drying and pyrolysis

LxWxH: 1.4x1.4x0.5 ± 1 mmMass: 0.061 grapp: 63 ± 20 kg m-3

sel: 5.5 – 55 S m-1

LxWxH: 1.8x1.8x0.8 ± 1 mmMass: 1.11 grapp: 428 ± 100 kg m-3

sel: 3 – 21 S m-1

130

131

La seconde partie de cette étude s’intéresse à l’analyse de la dégradation thermique des

précurseurs de carbone de MFC/LS ainsi qu’à leur cinétique de décomposition durant la

pyrolyse à haute température qui permet d’amorcer la formation de structure de graphite.

Au cours de cette étude les MFC et NaLS ont été caractérisés séparément et en mélanges en

termes de morphologie, de structure chimique et de stabilité thermique. Des différences

importantes ont été observées entre ces deux matériaux dans tous les domaines principalement

en raison de leur composition et de leur structure chimique différente. Selon la Figure 4a) et

b), les MFC se dégradent intensivement dans une plage de température étroite comprise entre

250 °C et 350 °C, tandis que le NaLS se décompose beaucoup plus doucement dans 2

gammes de températures principales, 150°C- 500°C et 600°C - 800°C.

Figure 4 : a) TG et b) DTG de MFC, NaLS (ou LS) et des mélanges (la flèche indique le

décalage du pic caractéristique de la cellulose ; c) Influence de la teneur en sodium sur ∆Tp

pour les mélanges MFC/LS.

c)

132

En outre, l'analyse cinétique des MFC et du NaLS a été menée avec deux approches basées

sur le modèle et sans modèle. Ce dernier suggère des paramètres d'Arrhenius trop élevés pour

être utilisés afin de reproduire la courbe expérimentale. Cependant, l'approche avec le modèle

permet de reproduire correctement la courbe expérimentale en fournissant des paramètres de

dégradation thermique des MFC tout à fait compatibles avec les données de la littérature

(Table. 1). L'analyse cinétique de la dégradation thermique de NaLS est réalisée ici pour la

première fois en fournissant un ensemble de données cinétiques qui permettent de prédire le

comportement thermique de ce matériau (Table. 2). De plus, les courbes DTG, qui

représentent la décomposition des composés MFC / NaLS mélangés dans divers rapports, ne

combinent pas "linéairement" les modèles des deux matériaux en raison de la présence de la

teneur en sodium. L'effet catalytique du sodium est confirmé par une relation fortement

corrélée entre ΔTp et la teneur en sodium (Figure 4c)) ainsi qu’expérimentalement par le

comportement thermique du mélange MFC/NaCl. D'un point de vue cinétique, une telle

diminution des températures de réaction pourrait être interprétée comme une diminution de

l'énergie d'activation des pseudo-composants de MFC dans les composés. En utilisant des

pseudo-composants moins élevé pour MFC, un fitting efficace a été atteint pour deux

composites MFC / NaLS en suivant une règle de superposition pondérée.

Table. 1: Parametres d’Arrhenius pour les 4 pseudo-composants de MFC

Pseudo-composants 1 2 3 4

Proportion 0.22 0.51 0.18 0.09

Ea (KJ/mol) 137 229 46 52

logA (log(s-1

)) 23.7 40.3 1.5 0.1

Table. 2 Parametres d’Arrhenius pour les 6 pseudo-composants de NaLS

Pseudo-composants 1 2 3 4 5 6

Proportion 0.05 0.17 0.19 0.15 0.24 0.20

Ea (KJ/mol) 149 112 132 104 45 72

LogA (log(s-1

)) 33 21 23 15 1.2 2

133

Une troisième étude est destinée à la caractérisation des bio-carbones de MFC/LS en

fonction de la température de la pyrolyse. En effet, cette dernière joue un rôle essentiel sur

certaines propriétés des carbones, telle que la densité, la microstructure et la conductivité

électrique.

Un large éventail de températures de traitement thermique (HTT de 400 à 1200 °C) a été

exploré pour la synthese des biocarbones afin de mieux comprendre l’évolution de leurs

propriétés morphologiques et microstructurales et, plus important, leurs propriétés électriques.

Les carbones MFC / LS se sont révélés résistants à la chaleur et morphologiquement stables à

la pyrolyse lente. Les résultats de FTIR sur les carbones MFC / LS ont fourni des éléments

tangibles concernant la perte de la majorité des groupes fonctionnels pendant le traitement

thermique. De plus, seulement quelques structures contenant de l'oxygène ont été retenues à

partir de la pyrolyse à haute température. La caractérisation microstructurale par Raman a

révélé que les carbones résultants sont composés de : i) des grappes de graphène (régulières)

sans défaut (3 couches de graphène, selon les mesures XRD); ii) des structures de graphène

avec défauts (principalement oxygène) et iii) des phases amorphes. Les résultats XRD

permettent de percevoir directement la croissance de grappes de graphène régulières avec des

HTT élevées tandis que les rapports de Raman (ID/Itotal, IG/Itotal et ID/IG) permettent de

comprendre les changements relatifs des trois phases. En raison de la croissance des phases

conductrices (régulières et avec défauts) en plus de la réduction des phases isolantes

(amorphe), une augmentation générale de la conductivité électrique a été observée dans les

carbones durs de MFC / LS en fonction des HTT bien que les mécanismes pour le

développement de la conductivité ne soit pas identiques pour les différentes gammes de

température (Figure 5):

i : 400 °C ≤ HTT ≤ 600 °C: les chars possèdent à cet état déjà des grappes de graphène

réguliers en petite taille (2.135 nm), comme en témoignent les résultats de XRD. Cependant,

ils sont complètement isolés par des structures de graphéne ‘défectueuses’ (détectées par la

bande de Raman D) et les structures amorphes (bande de Raman V, S et Gr). Cette dernière

est plus importante par rapport aux deux antérieures. Ainsi, les sauts d’électrons et l’effet

tunnel sont empêchés ou largement réduits. Par conséquent, la conductivité est indétectable ou

assez faible (Figure 5a)).

ii : 600 °C <HTT <700 °C: la phase amorphe continue de se réduire et participe ainsi au

développement des phases structurées. Atteignant un seuil de percolation, cettes dernières se

mettent en contact, ce qui favorise le mouvement des électrons en grande partie. Par

conséquent, la conductivité est nettement améliorée. Cependant, la phase de graphène

‘défectueuse’ étant beaucoup plus importante que celle de la phase de graphène régulière dans

les carbones MFC / LS (ce qui est attesté par le rapport Raman D / G toujours supérieur à 1),

on peut supposer que les voies électriques sont construites plus probablement par des contacts

entre les grappes de graphène défectueux (Figure 5b)).

134

iii : 700 °C ≤ HTT ≤ 1000 °C: pendant cette période, les structures de graphène régulières et

défectueuses continuent de croître en taille et en nombre, en intensifiant les contacts mutuels

entre les grappes conductrices. Par conséquent, la conductivité électrique est renforcée (Figure

5c)).

iv : HTTs> 1000 ° C: une fois la connexion électrique établie, une graphitisation renforcée ou

une réduction supplémentaire des défauts n'entraînera pas d’augmentation importante des

contacts entre les phases conductrices et un palier de la conductivité électrique induite par

HTT sera atteint (Figure 5d).

Malgré l’augmentation de la porosité pendant le traitement thermique (>400°C), les

conductivités apparentes des carbones MFC/LS sont fortement améliorées. Cet effet montre

bien que l’augmentation de la conductivité intrinseque de carbone domine sur le

développement de la porosité.

Après la carbonisation à 1000 ° C, les biocarbones des précurseurs de MFC / LS sont toujours

sous la forme de feuilles autoportées et peuvent présenter une conductivité électrique assez

intéressante (95 S / cm) par rapport à leur densité relativement faible de 1,14 g / cm3. Les

propriétés physiques mentionnées ci-dessus sont nettement supérieures par rapport aux

propriétés de biocarbones publiées dans la littérature.

Figure 5 : Modele permettant d’interpréter le développement de la conductivité électrique

dans les carbones MFC/LS.

135

La dernière étude de cette these résume les 3 études précédentes et examine l’utilisation de

mélanges de microfibrilles de cellulose / Lignosulfonate / poudre de cellulose référencés

MFC/LS/CP en tant que précurseur de carbone aussi bien en termes de comportement

rhéologique durant le procédé d’extrusion que de dégradation thermique durant la pyrolyse.

Les carbones issus des mélanges MFC/LS/CP sont alors caractérisés comme dans la section

précédente. La figure 6 synthétise l’ensemble des opérations réalisées dans cette section.

Figure 6 Utilisation de systèmes MFC/LS/CP comme précurseur de carbone : caractérisation

du comportement rhéologique jusqu’aux propriétés finales des carbones

Comme précédemment, des structures de carbone conductrices (électriquement) et résistantes

(mécaniquement) ont été élaborées par impression 3D puis pyrolysées. La processabilité des

‘slurry’ de MFC / LS / CP par l'impression en 3D a été examinée par des tests rhéologiques en

mode flow et thixotropique. Les objets imprimées sont autoportés avec une définition

d'impression élevée et se sont avérés morphologiquement stables au séchage à l'air ainsi

qu’apres l’opération de pyrolyse. Des mesures de TGA / DTG ont été effectuées pour

caractériser la dégradation thermique des composants individualisés ainsi que des mélanges

MFC / LS / CP. Les carbones résultants ont été ensuite caractérisés en termes de morphologie,

de microstructure et de propriétés physiques (comme la densité, la conductivité électrique et

la résistance mécanique). À 900 °C, les carbones MFC / LS / CP présentaient une

conductivité électrique élevée de 47,8 S / cm avec une faible densité de 0,74 g / cm3 ainsi

qu'une porosité importante de 58%. Ils ont également atteint un module d'élasticité maximal

de 6,62 GPa. Ces propriétés électriques et mécaniques peuvent ainsi conduire à utiliser dans

un proche avenir ces biocarbones dans des dispositifs de stockage d'énergie, en tant que

matériaux actifs d'électrode.

En conclusion, ces trois années de travail ont permis l’élaboration de structures de carbone

fonctionnelles qui pourraient être utilisées commet réseau conducteur dans les électrodes de

dispositif de stockage d’énergie. Grâce à une porosité contrôlable, une faible densité ainsi

qu’une conductivité électrique élevée, ce matériau apparait comme un candidat idéal pour la

fabrication d’électrodes structurées à haute capacité spécifique. En outre, ce travail propose

une nouvelle façon d’élaborer des électrodes en utilisant le procédé d’impression 3D. À

136

l’avenir, les électrodes pourraient être imprimées directement en utilisant une imprimant 3D à

deux têtes : l’une pour imprimer le matériau ‘d’accueil’ l’autre pour déposer des composants

actifs. Par ailleurs, ces travaux de these permettent de promouvoir l’utilisation de matériaux

lignocellulosiques dans les dispositifs de stockage d’énergie comme ‘matériaux actifs’.